-
Notifications
You must be signed in to change notification settings - Fork 0
/
featEng.py
173 lines (143 loc) · 5.13 KB
/
featEng.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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
__author__ = 'hujie'
import pandas as pd
import numpy as np
from sklearn import cross_validation
from sklearn.preprocessing import StandardScaler
from sklearn.utils import shuffle
import scipy.stats as stats
import random
from tqdm import tqdm
import gc
feat_size=26
#['bidder_id', 'auction', 'merchandise', 'device', 'time', 'country', 'ip', 'url']
# bidder_id,payment_account,address,outcome
random.seed()
def saveData(fpath,data,headers):
df = pd.DataFrame(data)
df.to_csv(fpath, header=headers,index=False)
print('Data has been saved!')
def sortByTime(bid):
return
def bidderFeatEng(bid_info):
'''
['bidder_id', 'auction', 'merchandise', 'device', 'time', 'country', 'ip', 'url']
return [
top_merchandise_1,top_merchandise_2,top_merchandise_3,unique_merchandise_count,
top_device_1,top_device_2,top_device_3,unique_device_count,
top_country_1,top_country_2,top_country_3,unique_country_count,
top_ip_1,top_ip_2,top_ip_3,unique_ip_count,
top_url_1,top_url_2,top_url_3,unique_url_count,
avg_time,var_time,zero_time,
avg_bids_per_auction, var_bids_per_auction, total_bids,
'''
# Transform list of lists into a numpy array
bid_info=np.array(bid_info)
count_list=[2,3,5,6,7]
rv=[]
for i in range(len(count_list)):
idx=count_list[i]
data=stats.itemfreq(bid_info[:,idx])
# Sort according to the count
data=data[data[:, 1].argsort()]
length=data.shape[0]
rv.append(data[0,0])
if length>=2:
rv.append(data[1,0])
else:
rv.append(0)
if length>=3:
rv.append(data[2,0])
else:
rv.append(0)
rv.append(length)
t = bid_info[:, 4].astype(float)
t_diff = t[1:] - t[:-1]
avg_time=0
var_time=0
zero_time=0
if t_diff.shape[0]>=1:
avg_time, var_time = t_diff.mean(), t_diff.var()
zero_time = t_diff.shape[0] - np.count_nonzero(t_diff)
rv.extend([avg_time,var_time,zero_time])
df = pd.DataFrame(bid_info[:, [0, 1]], columns=['bidder_id', 'auction'])
bids_count = df.groupby(['auction']).count()['bidder_id'].as_matrix()
avg_bids_per_auction, var_bids_per_auction= bids_count.mean(), bids_count.var()
total_bids=len(bid_info)
rv.extend([avg_bids_per_auction,var_bids_per_auction, total_bids])
return np.array(rv)
def generateData(fname,bidders,test=False):
data=pd.read_csv(fname)
data=data.values
data_x=np.zeros((len(data),feat_size))
data_y=np.zeros(len(data))
id=np.chararray(len(data),itemsize=37)
for i in tqdm(range(len(data))):
gc.collect()
bidder_name=data[i][0]
id[i]=bidder_name
if bidder_name in bidders:
bid_info=bidders[bidder_name]
data_x[i,:]=bidderFeatEng(bid_info)
'''
for j in range(min(len(bid_info),feat_size)):
idx = 7*j
for k in range(7):
data_x[i,idx+k]=bid_info[j][k+1]
'''
if not test:
data_y[i]=data[i][3]
if not test:
return data_x,data_y,id
else:
return data_x,id
def featEng():
bidData=pd.read_csv("./data/bids.csv")
bidData.set_index('bid_id',inplace = True)
bid=bidData.values
bidders={}
# mappings
map=[{},{},{},{},{},{},{},{}]
for i in range(len(bid)):
# Transforms data into integers
for j in range(1,8):
if bid[i][j] not in map[j]:
if j!=4:
map[j][bid[i][j]]=len(map[j])+1
else:
map[j][bid[i][j]]=bid[i][j]
bid[i][j]=map[j][bid[i][j]]
bidder_name=bid[i][0]
if bidder_name not in bidders:
bidders[bidder_name]=[]
bidders[bidder_name].append(bid[i])
for bidder_name in bidders:
bidders[bidder_name] = sorted(bidders[bidder_name], key=lambda bid: bid[4])
# Selects recent 100 bids for each bidder
test_x,test_id=generateData('./data/test.csv',bidders,test=True)
_train_x,_train_y,_train_id=generateData('./data/train.csv',bidders)
_train_x,_train_y,_train_id=shuffle(_train_x,_train_y,_train_id)
# 20% as validation data
# valid data: 0~valid_size-1
# train_size: ~len(_train_x)-1
valid_size=0.2*len(_train_x)
train_x=_train_x[valid_size+1:len(_train_x)-1,:]
train_y=_train_y[valid_size+1:len(_train_x)-1]
valid_x=_train_x[0:valid_size-1,:]
valid_y=_train_y[0:valid_size-1]
valid_id=_train_id[0:valid_size-1]
scaler = StandardScaler()
train_x = scaler.fit_transform(train_x)
valid_x = scaler.transform(valid_x)
test_x = scaler.transform(test_x)
x_headers=[]
for i in range(700):
x_headers.append("feat_"+str(i))
y_headers=["outcome"]
saveData("./data/train_x.csv",train_x,x_headers)
saveData("./data/train_y.csv",train_y,y_headers)
saveData("./data/valid_x.csv",valid_x,x_headers)
saveData("./data/valid_y.csv",valid_y,y_headers)
saveData("./data/test_x.csv",test_x,x_headers)
saveData("./data/valid_id.csv",valid_id,[])
saveData("./data/test_id.csv",test_id,[])
featEng()