-
Notifications
You must be signed in to change notification settings - Fork 5
/
uncertainty_analysis.py
25 lines (20 loc) · 1 KB
/
uncertainty_analysis.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
import numpy as np
import argparse
from experiment.uncertainty import uncertainty
from plots.rec_plots import show_uncertainty
from utils.io import load_numpy, save_dataframe_csv, find_best_hyperparameters, load_yaml
def main(args):
table_path = load_yaml('config/global.yml', key='path')['tables']
df = find_best_hyperparameters(table_path+args.problem, 'NDCG')
R_train = load_numpy(path=args.path, name=args.train)
results = uncertainty(R_train, df, rank=200)
show_uncertainty(results, x='numRated', y='std', hue='model',
folder=args.problem, name=args.name, save=True)
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="ParameterTuning")
parser.add_argument('-n', dest='name', default="uncertainty_analysis")
parser.add_argument('-d', dest='path', default="datax/")
parser.add_argument('-t', dest='train', default='Rtrain.npz')
parser.add_argument('-p', dest='problem', default='movielens1m')
args = parser.parse_args()
main(args)