|
a |
|
b/rf/args.py |
|
|
1 |
#!/usr/bin/env python |
|
|
2 |
|
|
|
3 |
from argparse import Namespace |
|
|
4 |
from pathlib import Path |
|
|
5 |
|
|
|
6 |
path = Path('../data') |
|
|
7 |
workdir=path/'workdir/rf' |
|
|
8 |
|
|
|
9 |
args = Namespace( |
|
|
10 |
workdir=workdir, |
|
|
11 |
dataset_csv=path/'proc_dataset.csv', |
|
|
12 |
cols=['hadm_id', 'imminent_adm_label', 'prolonged_stay_label', 'processed_note', 'charttime', 'intime', 'chartinterval'], |
|
|
13 |
imminent_adm_cols=['hadm_id', 'processed_note', 'imminent_adm_label'], |
|
|
14 |
prolonged_stay_cols=['hadm_id', 'processed_note', 'prolonged_stay_label'], |
|
|
15 |
dates=['charttime', 'intime'], |
|
|
16 |
modeldir=workdir/'models', |
|
|
17 |
min_freq=3, |
|
|
18 |
ia_thresh=0.27, |
|
|
19 |
ps_thresh=0.37, |
|
|
20 |
start_seed=127, |
|
|
21 |
) |
|
|
22 |
|
|
|
23 |
ia_params = { |
|
|
24 |
"class_weight": 'balanced', |
|
|
25 |
"criterion": "entropy", |
|
|
26 |
"max_features": 0.12, |
|
|
27 |
"min_samples_leaf": 3, |
|
|
28 |
"min_samples_split": 7, |
|
|
29 |
"n_estimators": 108, |
|
|
30 |
"oob_score": True, |
|
|
31 |
"n_jobs": 32, |
|
|
32 |
} |
|
|
33 |
|
|
|
34 |
ps_params = { |
|
|
35 |
"class_weight": 'balanced', |
|
|
36 |
"criterion": "entropy", |
|
|
37 |
"max_features": 0.114, |
|
|
38 |
"min_samples_leaf": 5, |
|
|
39 |
"min_samples_split": 3, |
|
|
40 |
"n_estimators": 116, |
|
|
41 |
"oob_score": True, |
|
|
42 |
"n_jobs": 32, |
|
|
43 |
} |