/
utils.py
163 lines (146 loc) · 5.26 KB
/
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
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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Date : 2019-07-25 13:13:14
# @Author : Mengji Zhang (zmj_xy@sjtu.edu.cn)
import numpy as np
import pandas as pd
import os
from six.moves import cPickle as pickle
import scipy.sparse as sp
def save_dict(di_,filename):
with open(filename,"wb") as f:
pickle.dump(di_,f)
def load_dict(filename):
with open(filename,"rb") as f:
ret_di=pickle.load(f)
return ret_di
def mkdirs(dirs):
if not os.path.exists(dirs):
os.makedirs(dirs)
def to_onehot(x,max=2):
onehot=np.zeros(shape=(max,))
onehot[x]=1
return onehot
def load_data(data_path,train_rate,val_rate):
df_org=pd.read_csv(data_path,header=0,engine="python")
df=df_org.dropna()
codes=sorted(list(set(df["code"].values)))
owners=sorted(list(set(df["S0802a"].values)))
codes_dict={code:i for i,code in enumerate(codes)}
owner_dict={owner:i for i,owner in enumerate(owners)}
xs=[]
ys=[]
for i in range(len(df)):
raw_x=df.iloc[i].values
x=[codes_dict[raw_x[0]],owner_dict[raw_x[2]],float(raw_x[4]),int(raw_x[7])-1]
y=to_onehot(raw_x[-1])
xs.append(x)
ys.append(y)
xs=np.array(xs)
ys=np.array(ys)
rng=np.random.RandomState(0)
all_idxs=np.arange(len(xs))
rng.shuffle(all_idxs)
train_idxs=all_idxs[:int(len(xs)*train_rate)]
val_idxs=all_idxs[int(len(xs)*train_rate):int(len(xs)*(train_rate+val_rate))]
test_idxs=all_idxs[int(len(xs)*(train_rate+val_rate)):]
train_xs=xs[train_idxs];train_ys=ys[train_idxs]
val_xs=xs[val_idxs];val_ys=ys[val_idxs]
test_xs=xs[test_idxs];test_ys=ys[test_idxs]
return train_xs,train_ys,val_xs,val_ys,test_xs,test_ys,codes_dict,owner_dict
def normalize(mx):
"""Row-normalize sparse matrix"""
rowsum = np.array(mx.sum(1))
r_inv = np.power(rowsum, -1).flatten()
r_inv[np.isinf(r_inv)] = 0.
r_mat_inv = sp.diags(r_inv)
mx = r_mat_inv.dot(mx)
return mx
def saved_data_for_graph(data_path,out_dir):
df_org=pd.read_csv(data_path,header=0,engine="python")
df=df_org.dropna()
codes=sorted(list(set(df["code"].values)))
owners=sorted(list(set(df["S0802a"].values)))
codes_dict={code:i for i,code in enumerate(codes)}
owner_dict={owner:i for i,owner in enumerate(owners)}
owner_adjacent=np.zeros(shape=(len(owners),len(owners)))
for i,owner in enumerate(owners):
print("{}/{}".format(i,len(owners)))
owner_df=df[df["S0802a"]==owner]
owned_codes=list(set(owner_df["code"].values))
for j,neighbor in enumerate(owners):
if neighbor==owner:
continue
neighbor_df=df[df["S0802a"]==neighbor]
neighbor_codes=list(set(neighbor_df["code"].values))
cnt=0
for o_code in owned_codes:
if o_code in neighbor_codes:
cnt+=1
weight=cnt/len(owned_codes)
owner_adjacent[i,j]=weight
if weight!=0:
print(weight,owned_codes,neighbor_codes)
# break
# break
owner_adjacent=normalize(owner_adjacent+sp.eye(owner_adjacent.shape[0]))
xs=[]
ys=[]
onwer_indexs=[]
for i in range(len(df)):
raw_x=df.iloc[i].values
x=[codes_dict[raw_x[0]],to_onehot(owner_dict[raw_x[2]],max=len(owners)),float(raw_x[4]),int(raw_x[7])-1]
y=to_onehot(raw_x[-1])
xs.append(x)
ys.append(y)
onwer_indexs.append(owner_dict[raw_x[2]])
onwer_indexs=np.array(onwer_indexs)
xs=np.array(xs)
ys=np.array(ys)
print(xs.sum())
print(ys.sum())
print(codes_dict["300339.SZ"])
print(owner_dict["陈强"])
print(owner_adjacent.sum())
print(onwer_indexs[100])
np.save(os.path.join(out_dir,"xs.npy"),xs)
np.save(os.path.join(out_dir,"ys.npy"),ys)
np.save(os.path.join(out_dir,"onwer_indexs.npy"),onwer_indexs)
save_dict(codes_dict,os.path.join(out_dir,"codes_dict.pkl"))
save_dict(owner_dict,os.path.join(out_dir,"owner_dict.pkl"))
np.save(os.path.join(out_dir,"owner_adjacent.npy"),owner_adjacent)
def load_data_for_graph(data_dir,train_rate,val_rate):
xs=np.load(os.path.join(data_dir,"xs.npy"),allow_pickle=True)
ys=np.load(os.path.join(data_dir,"ys.npy"))
onwer_indexs=np.load(os.path.join(data_dir,"onwer_indexs.npy"))
codes_dict=load_dict(os.path.join(data_dir,"codes_dict.pkl"))
owner_dict=load_dict(os.path.join(data_dir,"owner_dict.pkl"))
owner_adjacent=np.load(os.path.join(data_dir,"owner_adjacent.npy"))
owners_onehot=np.identity(len(owner_dict))
# print(xs.sum())
# print(ys.sum())
# print(codes_dict["300339.SZ"])
# print(owner_dict["陈强"])
# print(owner_adjacent.sum())
rng=np.random.RandomState(0)
all_idxs=np.arange(len(xs))
rng.shuffle(all_idxs)
train_idxs=all_idxs[:int(len(xs)*train_rate)]
val_idxs=all_idxs[int(len(xs)*train_rate):int(len(xs)*(train_rate+val_rate))]
test_idxs=all_idxs[int(len(xs)*(train_rate+val_rate)):]
train_xs=xs[train_idxs];train_ys=ys[train_idxs];train_owner_indexs=onwer_indexs[train_idxs]
val_xs=xs[val_idxs];val_ys=ys[val_idxs];val_owner_indexs=onwer_indexs[val_idxs]
test_xs=xs[test_idxs];test_ys=ys[test_idxs];test_owner_indexs=onwer_indexs[test_idxs]
return train_xs,train_owner_indexs,train_ys,val_xs,val_owner_indexs,val_ys,test_xs,test_owner_indexs,test_ys,codes_dict,owner_dict,owner_adjacent,owners_onehot
def compute_acc(pred,true):
preds=np.argmax(pred,axis=1)
trues=np.argmax(true,axis=1)
cnter=0
for p,t in zip(preds,trues):
if p==t:
cnter+=1
return cnter/len(pred)
if __name__=="__main__":
data_path="E:/deqin/processed_data/HLD_Chgequity_success_second_market2_clean.csv"
out_dir="E:/deqin/saved_data/";mkdirs(out_dir)
saved_data_for_graph(data_path,out_dir)