das3h/dataio.py at master · BenoitChoffin/das3h · GitHub

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
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
import os
import datetime
import numpy as np
from sklearn.model_selection import KFold

def prepare_folder(path):
	"""Create folder from path."""
	if not os.path.isdir(path):
		os.makedirs(path)

def build_new_paths(DATASET_NAME):
	"""Create dataset folder path name."""
	DATA_FOLDER = './data'
	CSV_FOLDER = os.path.join(DATA_FOLDER, DATASET_NAME)
	return CSV_FOLDER

def get_legend(experiment_args):
	"""Generate legend for an experiment.

	Argument:
	experiment_args -- experiment arguments (from ArgumentParser)

	Outputs:
	short -- short legend (str)
	full -- full legend (str)
	latex -- latex legend (str)
	active -- list of active variables
	"""
	dim = experiment_args['d']
	short = ''
	full = ''
	agents = ['users', 'items', 'skills', 'wins', 'fails', 'attempts']
	active = []
	for agent in agents:
		if experiment_args.get(agent):
			short += agent[0]
			active.append(agent)
	if experiment_args.get('tw_kc'):
		short += 't1'
		active.append("tw_kc")
	elif experiment_args.get('tw_items'):
		short += 't2'
		active.append("tw_items")
	short += '_' # add embedding dimension after underscore
	short += str(dim)
	prefix = ''
	if set(active) == {'users', 'items'} and dim == 0:
		prefix = 'IRT: '
	elif set(active) == {'users', 'items'} and dim > 0:
		prefix = 'MIRTb: '
	elif set(active) == {'skills', 'attempts'}:
		prefix = 'AFM: '
	elif set(active) == {'skills', 'wins', 'fails'}:
		prefix = 'PFA: '
	elif set(active) == {'users', 'items', 'skills', 'wins', 'attempts', 'tw_kc'}:
		prefix = 'DAS3H: '
	elif set(active) == {'users', 'items', 'wins', 'attempts', 'tw_items'}:
		prefix = 'DASH: '
	full = prefix + ', '.join(active) + ' d = {:d}'.format(dim)
	latex = prefix + ', '.join(active)
	return short, full, latex, active

def save_strongest_folds(full, dataset_name, nb_folds=5):
	all_users = full["user_id"].unique()
	kfold = KFold(nb_folds, shuffle=True)
	for i, (train, test) in enumerate(kfold.split(all_users)):
		path = "data/"+dataset_name+"/strongest/folds"
		prepare_folder(path)
		list_of_test_ids = []
		for user_id in test:
			list_of_test_ids += list(full.query('user_id == {}'.format(user_id)).index)
		np.save(path+'/test_fold{}.npy'.format(i), np.array(list_of_test_ids))

def save_pseudostrong_folds(full, dataset_name, perc_initial=.2, nb_folds=5):
	all_users = full["user_id"].unique()
	kfold = KFold(nb_folds, shuffle=True)
	for i, (train, test) in enumerate(kfold.split(all_users)):
		path = "data/"+dataset_name+"/pseudostrong/folds"
		prepare_folder(path)
		list_of_test_ids = []
		for user_id in test:
			fold = full.query('user_id == {}'.format(user_id)).sort_values('timestamp').index
			list_of_test_ids += list(fold[round(perc_initial*len(fold)):])
		np.save(path+'/test_fold{}.npy'.format(i), np.array(list_of_test_ids))