Mercurial > repos > bgruening > sklearn_lightgbm
annotate simple_model_fit.py @ 6:efe35ecda504 draft
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit f031d8ddfb73cec24572648666ac44ee47f08aad
| author | bgruening | 
|---|---|
| date | Thu, 11 Aug 2022 08:02:02 +0000 | 
| parents | 1555afbc31f6 | 
| children | 35e388240c33 | 
| rev | line source | 
|---|---|
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
1 import argparse | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
2 import json | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
3 import pickle | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
4 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
5 import pandas as pd | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
6 from galaxy_ml.utils import load_model, read_columns | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
7 from scipy.io import mmread | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
8 from sklearn.pipeline import Pipeline | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
9 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
10 N_JOBS = int(__import__("os").environ.get("GALAXY_SLOTS", 1)) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
11 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
12 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
13 # TODO import from galaxy_ml.utils in future versions | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
14 def clean_params(estimator, n_jobs=None): | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
15 """clean unwanted hyperparameter settings | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
16 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
17 If n_jobs is not None, set it into the estimator, if applicable | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
18 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
19 Return | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
20 ------ | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
21 Cleaned estimator object | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
22 """ | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
23 ALLOWED_CALLBACKS = ( | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
24 "EarlyStopping", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
25 "TerminateOnNaN", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
26 "ReduceLROnPlateau", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
27 "CSVLogger", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
28 "None", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
29 ) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
30 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
31 estimator_params = estimator.get_params() | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
32 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
33 for name, p in estimator_params.items(): | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
34 # all potential unauthorized file write | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
35 if name == "memory" or name.endswith("__memory") or name.endswith("_path"): | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
36 new_p = {name: None} | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
37 estimator.set_params(**new_p) | 
| 
3
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
38 elif n_jobs is not None and (name == "n_jobs" or name.endswith("__n_jobs")): | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
39 new_p = {name: n_jobs} | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
40 estimator.set_params(**new_p) | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
41 elif name.endswith("callbacks"): | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
42 for cb in p: | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
43 cb_type = cb["callback_selection"]["callback_type"] | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
44 if cb_type not in ALLOWED_CALLBACKS: | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
45 raise ValueError("Prohibited callback type: %s!" % cb_type) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
46 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
47 return estimator | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
48 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
49 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
50 def _get_X_y(params, infile1, infile2): | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
51 """read from inputs and output X and y | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
52 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
53 Parameters | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
54 ---------- | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
55 params : dict | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
56 Tool inputs parameter | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
57 infile1 : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
58 File path to dataset containing features | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
59 infile2 : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
60 File path to dataset containing target values | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
61 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
62 """ | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
63 # store read dataframe object | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
64 loaded_df = {} | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
65 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
66 input_type = params["input_options"]["selected_input"] | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
67 # tabular input | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
68 if input_type == "tabular": | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
69 header = "infer" if params["input_options"]["header1"] else None | 
| 
3
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
70 column_option = params["input_options"]["column_selector_options_1"][ | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
71 "selected_column_selector_option" | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
72 ] | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
73 if column_option in [ | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
74 "by_index_number", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
75 "all_but_by_index_number", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
76 "by_header_name", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
77 "all_but_by_header_name", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
78 ]: | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
79 c = params["input_options"]["column_selector_options_1"]["col1"] | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
80 else: | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
81 c = None | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
82 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
83 df_key = infile1 + repr(header) | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
84 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
85 loaded_df[df_key] = df | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
86 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
87 X = read_columns(df, c=c, c_option=column_option).astype(float) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
88 # sparse input | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
89 elif input_type == "sparse": | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
90 X = mmread(open(infile1, "r")) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
91 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
92 # Get target y | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
93 header = "infer" if params["input_options"]["header2"] else None | 
| 
3
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
94 column_option = params["input_options"]["column_selector_options_2"][ | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
95 "selected_column_selector_option2" | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
96 ] | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
97 if column_option in [ | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
98 "by_index_number", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
99 "all_but_by_index_number", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
100 "by_header_name", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
101 "all_but_by_header_name", | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
102 ]: | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
103 c = params["input_options"]["column_selector_options_2"]["col2"] | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
104 else: | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
105 c = None | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
106 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
107 df_key = infile2 + repr(header) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
108 if df_key in loaded_df: | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
109 infile2 = loaded_df[df_key] | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
110 else: | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
111 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
112 loaded_df[df_key] = infile2 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
113 | 
| 
3
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
114 y = read_columns( | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
115 infile2, c=c, c_option=column_option, sep="\t", header=header, parse_dates=True | 
| 
 
1555afbc31f6
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents: 
2 
diff
changeset
 | 
116 ) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
117 if len(y.shape) == 2 and y.shape[1] == 1: | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
118 y = y.ravel() | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
119 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
120 return X, y | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
121 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
122 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
123 def main(inputs, infile_estimator, infile1, infile2, out_object, out_weights=None): | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
124 """main | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
125 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
126 Parameters | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
127 ---------- | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
128 inputs : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
129 File path to galaxy tool parameter | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
130 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
131 infile_estimator : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
132 File paths of input estimator | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
133 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
134 infile1 : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
135 File path to dataset containing features | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
136 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
137 infile2 : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
138 File path to dataset containing target labels | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
139 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
140 out_object : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
141 File path for output of fitted model or skeleton | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
142 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
143 out_weights : str | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
144 File path for output of weights | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
145 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
146 """ | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
147 with open(inputs, "r") as param_handler: | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
148 params = json.load(param_handler) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
149 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
150 # load model | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
151 with open(infile_estimator, "rb") as est_handler: | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
152 estimator = load_model(est_handler) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
153 estimator = clean_params(estimator, n_jobs=N_JOBS) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
154 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
155 X_train, y_train = _get_X_y(params, infile1, infile2) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
156 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
157 estimator.fit(X_train, y_train) | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
158 | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
159 main_est = estimator | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
160 if isinstance(main_est, Pipeline): | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
161 main_est = main_est.steps[-1][-1] | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
162 if hasattr(main_est, "model_") and hasattr(main_est, "save_weights"): | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
163 if out_weights: | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
164 main_est.save_weights(out_weights) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
165 del main_est.model_ | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
166 del main_est.fit_params | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
167 del main_est.model_class_ | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
168 if getattr(main_est, "validation_data", None): | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
169 del main_est.validation_data | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
170 if getattr(main_est, "data_generator_", None): | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
171 del main_est.data_generator_ | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
172 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
173 with open(out_object, "wb") as output_handler: | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
174 pickle.dump(estimator, output_handler, pickle.HIGHEST_PROTOCOL) | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
175 | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
176 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
177 if __name__ == "__main__": | 
| 
0
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
178 aparser = argparse.ArgumentParser() | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
179 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
180 aparser.add_argument("-X", "--infile_estimator", dest="infile_estimator") | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
181 aparser.add_argument("-y", "--infile1", dest="infile1") | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
182 aparser.add_argument("-g", "--infile2", dest="infile2") | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
183 aparser.add_argument("-o", "--out_object", dest="out_object") | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
184 aparser.add_argument("-t", "--out_weights", dest="out_weights") | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
185 args = aparser.parse_args() | 
| 
 
dcbc668133e8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 756f8be9c3cd437e131e6410cd625c24fe078e8c"
 
bgruening 
parents:  
diff
changeset
 | 
186 | 
| 
2
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
187 main( | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
188 args.inputs, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
189 args.infile_estimator, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
190 args.infile1, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
191 args.infile2, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
192 args.out_object, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
193 args.out_weights, | 
| 
 
3c773ecadfdd
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
 
bgruening 
parents: 
0 
diff
changeset
 | 
194 ) | 
