-
Notifications
You must be signed in to change notification settings - Fork 24
/
data_utils.py
130 lines (113 loc) · 3.9 KB
/
data_utils.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
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
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
import argparse
import os
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder, MinMaxScaler
sparse_features = {
"criteo_kaggle": ["C" + str(i) for i in range(1, 27)],
"avazu": [
"hour",
"C1",
"banner_pos",
"site_id",
"site_domain",
"site_category",
"app_id",
"app_domain",
"app_category",
"device_id",
"device_ip",
"device_model",
"device_type",
"device_conn_type",
"C14",
"C15",
"C16",
"C17",
"C18",
"C19",
"C20",
"C21",
]
}
dense_features = {
"criteo_kaggle": ["I" + str(i) for i in range(1, 14)],
"avazu": []
}
target = {
"criteo_kaggle": ["label"],
"avazu": ["click"],
}
data_path = {
"criteo_kaggle": "data/criteo_kaggle",
"avazu": "data/avazu",
}
split_rand_ratio = {
"criteo_kaggle": 0.1,
"avazu": 0.2
}
def load_data(dataset="criteo_kaggle", split="rand"):
train_file = os.path.join(
data_path[dataset], f"{dataset}_processed_{split}_train.feather")
test_file = os.path.join(
data_path[dataset], f"{dataset}_processed_{split}_test.feather")
train_data = pd.read_feather(train_file)
test_data = pd.read_feather(test_file)
return train_data, test_data
def load_feature_name(dataset="criteo_kaggle"):
return sparse_features[dataset], dense_features[dataset], target[dataset]
def preprocess(data, sparse_features, dense_features):
data[sparse_features] = data[sparse_features].fillna(
"-1",
)
for feat in sparse_features:
lbe = LabelEncoder()
data[feat] = lbe.fit_transform(data[feat])
if len(dense_features) > 0:
data[dense_features] = data[dense_features].fillna(
0,
)
mms = MinMaxScaler(feature_range=(0, 1))
data[dense_features] = mms.fit_transform(data[dense_features])
return data
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--dataset", default="criteo_kaggle",
choices=["criteo_kaggle", "criteo_terabyte", "avazu", "taobao"], type=str)
parser.add_argument("--split", default=None,
choices=["seq", "rand"], type=str)
args = parser.parse_args()
o_file = os.path.join(data_path[args.dataset], f"{args.dataset}_processed")
if args.dataset == "criteo_kaggle":
i_file = os.path.join(data_path[args.dataset], "train.txt")
data = pd.read_csv(i_file, delimiter="\t", header=None)
data.columns = pd.read_csv(os.path.join(
data_path[args.dataset], "criteo_sample.txt")).columns
elif args.dataset == "avazu":
i_file = os.path.join(data_path[args.dataset], "train")
data = pd.read_csv(i_file, dtype={"id": str})
sparse_features, dense_features, target = load_feature_name(args.dataset)
data = preprocess(data, sparse_features, dense_features)
if args.split is not None:
if args.split == "seq":
if args.dataset == "criteo_kaggle":
test_ratio = 1/7
train_size = int((1-test_ratio) * len(data))
train = data.iloc[:train_size]
test = data.iloc[train_size:]
else:
raise NotImplementedError
else:
data_index = np.arange(len(data))
test_ratio = split_rand_ratio[args.dataset]
train_index, test_index = train_test_split(
data_index, test_size=test_ratio)
train = data.iloc[train_index]
test = data.iloc[test_index]
train.reset_index(drop=True).to_feather(
o_file + f"_{args.split}_train.feather")
test.reset_index(drop=True).to_feather(
o_file + f"_{args.split}_test.feather")
else:
data.to_feather(o_file + ".feather")