Mercurial > repos > goeckslab > ludwig_config_generator
annotate ludwig_autogenconfig.py @ 4:36c5fcc49286 draft default tip
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit 4dc221b2fa9717552787f0985ad3fc3df4460158
author | goeckslab |
---|---|
date | Sat, 21 Jun 2025 15:05:41 +0000 |
parents | 183adfc24076 |
children |
rev | line source |
---|---|
0
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
1 import argparse |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
2 import logging |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
3 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
4 from ludwig import automl |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
5 from ludwig.utils import defaults |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
6 from pandas import read_csv |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
7 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
8 logging.basicConfig(level=logging.DEBUG) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
9 LOG = logging.getLogger(__name__) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
10 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
11 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
12 def main(): |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
13 parser = argparse.ArgumentParser( |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
14 description='Render a Ludwig config') |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
15 parser.add_argument( |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
16 '--dataset', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
17 type=str, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
18 help='Path to the dataset file', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
19 required=True) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
20 parser.add_argument( |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
21 '--output_feature', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
22 type=int, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
23 help='Name for the output feature', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
24 required=True) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
25 parser.add_argument( |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
26 '--output', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
27 type=str, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
28 help='Path for the output file', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
29 required=True) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
30 parser.add_argument( |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
31 '--renderconfig', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
32 action='store_true', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
33 help='Render the config', |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
34 required=False, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
35 default=False) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
36 args = parser.parse_args() |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
37 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
38 # get the output feature name |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
39 df = read_csv(args.dataset, nrows=2, sep=None, engine='python') |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
40 names = df.columns.tolist() |
4
36c5fcc49286
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit 4dc221b2fa9717552787f0985ad3fc3df4460158
goeckslab
parents:
0
diff
changeset
|
41 target = names[args.output_feature - 1] |
0
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
42 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
43 args_init = ["--dataset", args.dataset, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
44 "--target", target, |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
45 "--output", args.output] |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
46 automl.cli_init_config(args_init) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
47 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
48 if args.renderconfig: |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
49 args_render = ["--config", args.output, "--output", args.output] |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
50 defaults.cli_render_config(args_render) |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
51 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
52 |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
53 if __name__ == "__main__": |
183adfc24076
planemo upload for repository https://github.com/goeckslab/Galaxy-Ludwig.git commit bdea9430787658783a51cc6c2ae951a01e455bb4
goeckslab
parents:
diff
changeset
|
54 main() |