Mercurial > repos > bgruening > sklearn_to_categorical

annotate keras_train_and_eval.py @ 7:3f088677e4ff draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
author bgruening
date Mon, 02 Oct 2023 07:56:37 +0000
parents aa77f760a04e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
1 import argparse
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
2 import json
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
3 import os
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
4 import warnings
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
5 from itertools import chain
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
6
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
7 import joblib
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
8 import numpy as np
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
9 import pandas as pd
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
10 from galaxy_ml.keras_galaxy_models import (
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
11 _predict_generator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
12 KerasGBatchClassifier,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
13 )
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
14 from galaxy_ml.model_persist import dump_model_to_h5, load_model_from_h5
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
15 from galaxy_ml.model_validations import train_test_split
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
16 from galaxy_ml.utils import (
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
17 clean_params,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
18 gen_compute_scores,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
19 get_main_estimator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
20 get_module,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
21 get_scoring,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
22 read_columns,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
23 SafeEval
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
24 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
25 from scipy.io import mmread
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
26 from sklearn.metrics._scorer import _check_multimetric_scoring
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
27 from sklearn.model_selection._validation import _score
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
28 from sklearn.utils import _safe_indexing, indexable
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
29
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
30 N_JOBS = int(os.environ.get("GALAXY_SLOTS", 1))
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
31 CACHE_DIR = os.path.join(os.getcwd(), "cached")
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
32 NON_SEARCHABLE = (
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
33 "n_jobs",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
34 "pre_dispatch",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
35 "memory",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
36 "_path",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
37 "_dir",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
38 "nthread",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
39 "callbacks",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
40 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
41 ALLOWED_CALLBACKS = (
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
42 "EarlyStopping",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
43 "TerminateOnNaN",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
44 "ReduceLROnPlateau",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
45 "CSVLogger",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
46 "None",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
47 )
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
48
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
49
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
50 def _eval_swap_params(params_builder):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
51 swap_params = {}
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
52
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
53 for p in params_builder["param_set"]:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
54 swap_value = p["sp_value"].strip()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
55 if swap_value == "":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
56 continue
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
57
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
58 param_name = p["sp_name"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
59 if param_name.lower().endswith(NON_SEARCHABLE):
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
60 warnings.warn(
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
61 "Warning: `%s` is not eligible for search and was "
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
62 "omitted!" % param_name
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
63 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
64 continue
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
65
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
66 if not swap_value.startswith(":"):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
67 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
68 ev = safe_eval(swap_value)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
69 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
70 # Have `:` before search list, asks for estimator evaluatio
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
71 safe_eval_es = SafeEval(load_estimators=True)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
72 swap_value = swap_value[1:].strip()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
73 # TODO maybe add regular express check
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
74 ev = safe_eval_es(swap_value)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
75
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
76 swap_params[param_name] = ev
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
77
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
78 return swap_params
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
79
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
80
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
81 def train_test_split_none(*arrays, **kwargs):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
82 """extend train_test_split to take None arrays
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
83 and support split by group names.
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
84 """
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
85 nones = []
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
86 new_arrays = []
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
87 for idx, arr in enumerate(arrays):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
88 if arr is None:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
89 nones.append(idx)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
90 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
91 new_arrays.append(arr)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
92
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
93 if kwargs["shuffle"] == "None":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
94 kwargs["shuffle"] = None
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
95
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
96 group_names = kwargs.pop("group_names", None)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
97
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
98 if group_names is not None and group_names.strip():
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
99 group_names = [name.strip() for name in group_names.split(",")]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
100 new_arrays = indexable(*new_arrays)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
101 groups = kwargs["labels"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
102 n_samples = new_arrays[0].shape[0]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
103 index_arr = np.arange(n_samples)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
104 test = index_arr[np.isin(groups, group_names)]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
105 train = index_arr[~np.isin(groups, group_names)]
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
106 rval = list(
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
107 chain.from_iterable(
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
108 (_safe_indexing(a, train), _safe_indexing(a, test)) for a in new_arrays
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
109 )
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
110 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
111 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
112 rval = train_test_split(*new_arrays, **kwargs)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
113
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
114 for pos in nones:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
115 rval[pos * 2: 2] = [None, None]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
116
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
117 return rval
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
118
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
119
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
120 def _evaluate_keras_and_sklearn_scores(
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
121 estimator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
122 data_generator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
123 X,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
124 y=None,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
125 sk_scoring=None,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
126 steps=None,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
127 batch_size=32,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
128 return_predictions=False,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
129 ):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
130 """output scores for bother keras and sklearn metrics
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
131
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
132 Parameters
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
133 -----------
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
134 estimator : object
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
135 Fitted `galaxy_ml.keras_galaxy_models.KerasGBatchClassifier`.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
136 data_generator : object
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
137 From `galaxy_ml.preprocessors.ImageDataFrameBatchGenerator`.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
138 X : 2-D array
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
139 Contains indecies of images that need to be evaluated.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
140 y : None
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
141 Target value.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
142 sk_scoring : dict
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
143 Galaxy tool input parameters.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
144 steps : integer or None
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
145 Evaluation/prediction steps before stop.
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
146 batch_size : integer
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
147 Number of samples in a batch
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
148 return_predictions : bool, default is False
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
149 Whether to return predictions and true labels.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
150 """
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
151 scores = {}
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
152
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
153 generator = data_generator.flow(X, y=y, batch_size=batch_size)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
154 # keras metrics evaluation
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
155 # handle scorer, convert to scorer dict
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
156 generator.reset()
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
157 score_results = estimator.model_.evaluate_generator(generator, steps=steps)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
158 metrics_names = estimator.model_.metrics_names
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
159 if not isinstance(metrics_names, list):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
160 scores[metrics_names] = score_results
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
161 else:
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
162 scores = dict(zip(metrics_names, score_results))
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
163
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
164 if sk_scoring["primary_scoring"] == "default" and not return_predictions:
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
165 return scores
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
166
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
167 generator.reset()
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
168 predictions, y_true = _predict_generator(estimator.model_, generator, steps=steps)
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
169
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
170 # for sklearn metrics
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
171 if sk_scoring["primary_scoring"] != "default":
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
172 scorer = get_scoring(sk_scoring)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
173 if not isinstance(scorer, (dict, list)):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
174 scorer = [sk_scoring["primary_scoring"]]
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
175 scorer = _check_multimetric_scoring(estimator, scoring=scorer)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
176 sk_scores = gen_compute_scores(y_true, predictions, scorer)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
177 scores.update(sk_scores)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
178
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
179 if return_predictions:
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
180 return scores, predictions, y_true
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
181 else:
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
182 return scores, None, None
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
183
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
184
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
185 def main(
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
186 inputs,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
187 infile_estimator,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
188 infile1,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
189 infile2,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
190 outfile_result,
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
191 outfile_history=None,
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
192 outfile_object=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
193 outfile_y_true=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
194 outfile_y_preds=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
195 groups=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
196 ref_seq=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
197 intervals=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
198 targets=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
199 fasta_path=None,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
200 ):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
201 """
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
202 Parameter
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
203 ---------
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
204 inputs : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
205 File path to galaxy tool parameter.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
206
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
207 infile_estimator : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
208 File path to estimator.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
209
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
210 infile1 : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
211 File path to dataset containing features.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
212
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
213 infile2 : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
214 File path to dataset containing target values.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
215
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
216 outfile_result : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
217 File path to save the results, either cv_results or test result.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
218
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
219 outfile_history : str, optional
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
220 File path to save the training history.
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
221
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
222 outfile_object : str, optional
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
223 File path to save searchCV object.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
224
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
225 outfile_y_true : str, optional
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
226 File path to target values for prediction.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
227
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
228 outfile_y_preds : str, optional
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
229 File path to save predictions.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
230
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
231 groups : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
232 File path to dataset containing groups labels.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
233
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
234 ref_seq : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
235 File path to dataset containing genome sequence file.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
236
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
237 intervals : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
238 File path to dataset containing interval file.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
239
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
240 targets : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
241 File path to dataset compressed target bed file.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
242
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
243 fasta_path : str
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
244 File path to dataset containing fasta file.
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
245 """
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
246 warnings.simplefilter("ignore")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
247
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
248 with open(inputs, "r") as param_handler:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
249 params = json.load(param_handler)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
250
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
251 # load estimator
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
252 estimator = load_model_from_h5(infile_estimator)
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
253
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
254 estimator = clean_params(estimator)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
255
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
256 # swap hyperparameter
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
257 swapping = params["experiment_schemes"]["hyperparams_swapping"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
258 swap_params = _eval_swap_params(swapping)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
259 estimator.set_params(**swap_params)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
260 estimator_params = estimator.get_params()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
261 # store read dataframe object
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
262 loaded_df = {}
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
263
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
264 input_type = params["input_options"]["selected_input"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
265 # tabular input
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
266 if input_type == "tabular":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
267 header = "infer" if params["input_options"]["header1"] else None
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
268 column_option = params["input_options"]["column_selector_options_1"][
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
269 "selected_column_selector_option"
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
270 ]
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
271 if column_option in [
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
272 "by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
273 "all_but_by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
274 "by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
275 "all_but_by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
276 ]:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
277 c = params["input_options"]["column_selector_options_1"]["col1"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
278 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
279 c = None
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
280
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
281 df_key = infile1 + repr(header)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
282 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
283 loaded_df[df_key] = df
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
284
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
285 X = read_columns(df, c=c, c_option=column_option).astype(float)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
286 # sparse input
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
287 elif input_type == "sparse":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
288 X = mmread(open(infile1, "r"))
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
289
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
290 # fasta_file input
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
291 elif input_type == "seq_fasta":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
292 pyfaidx = get_module("pyfaidx")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
293 sequences = pyfaidx.Fasta(fasta_path)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
294 n_seqs = len(sequences.keys())
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
295 X = np.arange(n_seqs)[:, np.newaxis]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
296 for param in estimator_params.keys():
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
297 if param.endswith("fasta_path"):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
298 estimator.set_params(**{param: fasta_path})
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
299 break
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
300 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
301 raise ValueError(
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
302 "The selected estimator doesn't support "
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
303 "fasta file input! Please consider using "
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
304 "KerasGBatchClassifier with "
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
305 "FastaDNABatchGenerator/FastaProteinBatchGenerator "
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
306 "or having GenomeOneHotEncoder/ProteinOneHotEncoder "
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
307 "in pipeline!"
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
308 )
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
309
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
310 elif input_type == "refseq_and_interval":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
311 path_params = {
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
312 "data_batch_generator__ref_genome_path": ref_seq,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
313 "data_batch_generator__intervals_path": intervals,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
314 "data_batch_generator__target_path": targets,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
315 }
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
316 estimator.set_params(**path_params)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
317 n_intervals = sum(1 for line in open(intervals))
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
318 X = np.arange(n_intervals)[:, np.newaxis]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
319
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
320 # Get target y
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
321 header = "infer" if params["input_options"]["header2"] else None
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
322 column_option = params["input_options"]["column_selector_options_2"][
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
323 "selected_column_selector_option2"
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
324 ]
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
325 if column_option in [
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
326 "by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
327 "all_but_by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
328 "by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
329 "all_but_by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
330 ]:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
331 c = params["input_options"]["column_selector_options_2"]["col2"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
332 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
333 c = None
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
334
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
335 df_key = infile2 + repr(header)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
336 if df_key in loaded_df:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
337 infile2 = loaded_df[df_key]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
338 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
339 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
340 loaded_df[df_key] = infile2
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
341
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
342 y = read_columns(
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
343 infile2,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
344 c=c,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
345 c_option=column_option,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
346 sep="\t",
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
347 header=header,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
348 parse_dates=True,
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
349 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
350 if len(y.shape) == 2 and y.shape[1] == 1:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
351 y = y.ravel()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
352 if input_type == "refseq_and_interval":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
353 estimator.set_params(data_batch_generator__features=y.ravel().tolist())
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
354 y = None
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
355 # end y
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
356
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
357 # load groups
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
358 if groups:
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
359 groups_selector = (
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
360 params["experiment_schemes"]["test_split"]["split_algos"]
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
361 ).pop("groups_selector")
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
362
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
363 header = "infer" if groups_selector["header_g"] else None
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
364 column_option = groups_selector["column_selector_options_g"][
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
365 "selected_column_selector_option_g"
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
366 ]
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
367 if column_option in [
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
368 "by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
369 "all_but_by_index_number",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
370 "by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
371 "all_but_by_header_name",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
372 ]:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
373 c = groups_selector["column_selector_options_g"]["col_g"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
374 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
375 c = None
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
376
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
377 df_key = groups + repr(header)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
378 if df_key in loaded_df:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
379 groups = loaded_df[df_key]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
380
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
381 groups = read_columns(
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
382 groups,
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
383 c=c,
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
384 c_option=column_option,
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
385 sep="\t",
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
386 header=header,
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
387 parse_dates=True,
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
388 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
389 groups = groups.ravel()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
390
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
391 # del loaded_df
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
392 del loaded_df
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
393
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
394 # cache iraps_core fits could increase search speed significantly
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
395 memory = joblib.Memory(location=CACHE_DIR, verbose=0)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
396 main_est = get_main_estimator(estimator)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
397 if main_est.__class__.__name__ == "IRAPSClassifier":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
398 main_est.set_params(memory=memory)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
399
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
400 # handle scorer, convert to scorer dict
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
401 scoring = params["experiment_schemes"]["metrics"]["scoring"]
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
402 scorer = get_scoring(scoring)
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
403 if not isinstance(scorer, (dict, list)):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
404 scorer = [scoring["primary_scoring"]]
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
405 scorer = _check_multimetric_scoring(estimator, scoring=scorer)
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
406
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
407 # handle test (first) split
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
408 test_split_options = params["experiment_schemes"]["test_split"]["split_algos"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
409
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
410 if test_split_options["shuffle"] == "group":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
411 test_split_options["labels"] = groups
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
412 if test_split_options["shuffle"] == "stratified":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
413 if y is not None:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
414 test_split_options["labels"] = y
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
415 else:
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
416 raise ValueError(
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
417 "Stratified shuffle split is not " "applicable on empty target values!"
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
418 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
419
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
420 X_train, X_test, y_train, y_test, groups_train, groups_test = train_test_split_none(
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
421 X, y, groups, **test_split_options
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
422 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
423
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
424 exp_scheme = params["experiment_schemes"]["selected_exp_scheme"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
425
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
426 # handle validation (second) split
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
427 if exp_scheme == "train_val_test":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
428 val_split_options = params["experiment_schemes"]["val_split"]["split_algos"]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
429
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
430 if val_split_options["shuffle"] == "group":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
431 val_split_options["labels"] = groups_train
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
432 if val_split_options["shuffle"] == "stratified":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
433 if y_train is not None:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
434 val_split_options["labels"] = y_train
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
435 else:
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
436 raise ValueError(
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
437 "Stratified shuffle split is not "
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
438 "applicable on empty target values!"
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
439 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
440
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
441 (
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
442 X_train,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
443 X_val,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
444 y_train,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
445 y_val,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
446 groups_train,
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
447 groups_val,
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
448 ) = train_test_split_none(X_train, y_train, groups_train, **val_split_options)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
449
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
450 # train and eval
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
451 if hasattr(estimator, "config") and hasattr(estimator, "model_type"):
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
452 if exp_scheme == "train_val_test":
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
453 history = estimator.fit(X_train, y_train, validation_data=(X_val, y_val))
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
454 else:
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
455 history = estimator.fit(X_train, y_train, validation_data=(X_test, y_test))
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
456 else:
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
457 history = estimator.fit(X_train, y_train)
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
458 if "callbacks" in estimator_params:
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
459 for cb in estimator_params["callbacks"]:
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
460 if cb["callback_selection"]["callback_type"] == "CSVLogger":
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
461 hist_df = pd.DataFrame(history.history)
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
462 hist_df["epoch"] = np.arange(1, estimator_params["epochs"] + 1)
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
463 epo_col = hist_df.pop('epoch')
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
464 hist_df.insert(0, 'epoch', epo_col)
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
465 hist_df.to_csv(path_or_buf=outfile_history, sep="\t", header=True, index=False)
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
466 break
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
467 if isinstance(estimator, KerasGBatchClassifier):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
468 scores = {}
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
469 steps = estimator.prediction_steps
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
470 batch_size = estimator.batch_size
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
471 data_generator = estimator.data_generator_
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
472
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
473 scores, predictions, y_true = _evaluate_keras_and_sklearn_scores(
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
474 estimator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
475 data_generator,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
476 X_test,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
477 y=y_test,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
478 sk_scoring=scoring,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
479 steps=steps,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
480 batch_size=batch_size,
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
481 return_predictions=bool(outfile_y_true),
1
2cb67aeee0d9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 0
diff changeset
482 )
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
483
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
484 else:
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
485 scores = {}
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
486 if hasattr(estimator, "model_") and hasattr(estimator.model_, "metrics_names"):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
487 batch_size = estimator.batch_size
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
488 score_results = estimator.model_.evaluate(
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
489 X_test, y=y_test, batch_size=batch_size, verbose=0
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
490 )
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
491 metrics_names = estimator.model_.metrics_names
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
492 if not isinstance(metrics_names, list):
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
493 scores[metrics_names] = score_results
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
494 else:
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
495 scores = dict(zip(metrics_names, score_results))
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
496
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
497 if hasattr(estimator, "predict_proba"):
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
498 predictions = estimator.predict_proba(X_test)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
499 else:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
500 predictions = estimator.predict(X_test)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
501
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
502 y_true = y_test
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
503 sk_scores = _score(estimator, X_test, y_test, scorer)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
504 scores.update(sk_scores)
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
505
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
506 # handle output
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
507 if outfile_y_true:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
508 try:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
509 pd.DataFrame(y_true).to_csv(outfile_y_true, sep="\t", index=False)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
510 pd.DataFrame(predictions).astype(np.float32).to_csv(
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
511 outfile_y_preds,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
512 sep="\t",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
513 index=False,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
514 float_format="%g",
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
515 chunksize=10000,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
516 )
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
517 except Exception as e:
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
518 print("Error in saving predictions: %s" % e)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
519 # handle output
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
520 for name, score in scores.items():
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
521 scores[name] = [score]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
522 df = pd.DataFrame(scores)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
523 df = df[sorted(df.columns)]
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
524 df.to_csv(path_or_buf=outfile_result, sep="\t", header=True, index=False)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
525
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
526 memory.clear(warn=False)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
527
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
528 if outfile_object:
5
aa77f760a04e planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 1
diff changeset
529 dump_model_to_h5(estimator, outfile_object)
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
530
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
531
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
532 if __name__ == "__main__":
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
533 aparser = argparse.ArgumentParser()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
534 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
535 aparser.add_argument("-e", "--estimator", dest="infile_estimator")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
536 aparser.add_argument("-X", "--infile1", dest="infile1")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
537 aparser.add_argument("-y", "--infile2", dest="infile2")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
538 aparser.add_argument("-O", "--outfile_result", dest="outfile_result")
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
539 aparser.add_argument("-hi", "--outfile_history", dest="outfile_history")
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
540 aparser.add_argument("-o", "--outfile_object", dest="outfile_object")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
541 aparser.add_argument("-l", "--outfile_y_true", dest="outfile_y_true")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
542 aparser.add_argument("-p", "--outfile_y_preds", dest="outfile_y_preds")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
543 aparser.add_argument("-g", "--groups", dest="groups")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
544 aparser.add_argument("-r", "--ref_seq", dest="ref_seq")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
545 aparser.add_argument("-b", "--intervals", dest="intervals")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
546 aparser.add_argument("-t", "--targets", dest="targets")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
547 aparser.add_argument("-f", "--fasta_path", dest="fasta_path")
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
548 args = aparser.parse_args()
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
549
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
550 main(
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
551 args.inputs,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
552 args.infile_estimator,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
553 args.infile1,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
554 args.infile2,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
555 args.outfile_result,
7
3f088677e4ff planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a
bgruening
parents: 5
diff changeset
556 outfile_history=args.outfile_history,
0
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
557 outfile_object=args.outfile_object,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
558 outfile_y_true=args.outfile_y_true,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
559 outfile_y_preds=args.outfile_y_preds,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
560 groups=args.groups,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
561 ref_seq=args.ref_seq,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
562 intervals=args.intervals,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
563 targets=args.targets,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
564 fasta_path=args.fasta_path,
bdf3f88c60e0 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents:
diff changeset
565 )