forked from guoday/Tencent2020_Rank1st
-
Notifications
You must be signed in to change notification settings - Fork 0
/
run.py
160 lines (146 loc) · 7.4 KB
/
run.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
import os
import gc
import torch
import logging
import argparse
import models.ctrNet as ctrNet
import pickle
import gensim
import random
import pandas as pd
import numpy as np
from tqdm import tqdm
from src.data_loader import TextDataset
from torch.utils.data import DataLoader, SequentialSampler, RandomSampler,TensorDataset
from sklearn.preprocessing import LabelEncoder, MinMaxScaler, StandardScaler
from sklearn.model_selection import StratifiedKFold
base_path="data"
#定义浮点数特征
dense_features=['user_id__size', 'user_id_ad_id_unique', 'user_id_creative_id_unique', 'user_id_advertiser_id_unique', 'user_id_industry_unique', 'user_id_product_id_unique', 'user_id_time_unique', 'user_id_click_times_sum', 'user_id_click_times_mean', 'user_id_click_times_std']
for l in ['age_{}'.format(i) for i in range(10)]+['gender_{}'.format(i) for i in range(2)]:
for f in ['creative_id','ad_id','product_id','advertiser_id','industry']:
dense_features.append(l+'_'+f+'_mean')
#定义用户点击的序列特征
text_features=[
[base_path+"/sequence_text_user_id_product_id.128d",'sequence_text_user_id_product_id',128],
[base_path+"/sequence_text_user_id_ad_id.128d",'sequence_text_user_id_ad_id',128],
[base_path+"/sequence_text_user_id_creative_id.128d",'sequence_text_user_id_creative_id',128],
[base_path+"/sequence_text_user_id_advertiser_id.128d",'sequence_text_user_id_advertiser_id',128],
[base_path+"/sequence_text_user_id_industry.128d",'sequence_text_user_id_industry',128],
[base_path+"/sequence_text_user_id_product_category.128d",'sequence_text_user_id_product_category',128],
[base_path+"/sequence_text_user_id_time.128d",'sequence_text_user_id_time',128],
[base_path+"/sequence_text_user_id_click_times.128d",'sequence_text_user_id_click_times',128],
]
#定义用户点击的人工构造序列特征
text_features_1=[
[base_path+"/sequence_text_user_id_creative_id_fold.12d",'sequence_text_user_id_creative_id_fold',12],
[base_path+"/sequence_text_user_id_ad_id_fold.12d",'sequence_text_user_id_ad_id_fold',12],
[base_path+"/sequence_text_user_id_product_id_fold.12d",'sequence_text_user_id_product_id_fold',12],
[base_path+"/sequence_text_user_id_advertiser_id_fold.12d",'sequence_text_user_id_advertiser_id_fold',12],
[base_path+"/sequence_text_user_id_industry_fold.12d",'sequence_text_user_id_industry_fold',12],
]
if __name__ == "__main__":
logger = logging.getLogger(__name__)
logging.basicConfig(format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',
datefmt='%m/%d/%Y %H:%M:%S',
level=logging.INFO)
parser = argparse.ArgumentParser()
parser.add_argument('--kfold', type=int, default=5)
parser.add_argument('--index', type=int, default=0)
parser.add_argument('--train_batch_size', type=int, default=512)
parser.add_argument('--max_len_text', type=int, default=128)
parser.add_argument('--num_hidden_layers', type=int, default=6)
parser.add_argument('--hidden_dropout_prob', type=float, default=0.2)
parser.add_argument('--output_path', type=str, default=None)
parser.add_argument('--pretrained_model_path', type=str, default=None)
parser.add_argument('--hidden_size', type=int, default=1024)
parser.add_argument('--vocab_size_v1', type=int, default=500000)
parser.add_argument('--vocab_dim_v1', type=int, default=64)
parser.add_argument('--epoch', type=int, default=5)
parser.add_argument('--lr', type=float, default=8e-5)
parser.add_argument('--eval_steps', type=int, default=500)
parser.add_argument('--display_steps', type=int, default=100)
parser.add_argument('--max_grad_norm', type=float, default=1.0)
parser.add_argument('--eval_batch_size', type=int, default=4096)
parser.add_argument('--seed', type=int, default=2020)
parser.add_argument('--num_label', type=int, default=20)
args = parser.parse_args()
#设置参数
args.hidden_size=sum([x[-1] for x in text_features])
logger.info("Argument %s", args)
args.vocab=pickle.load(open(os.path.join(args.pretrained_model_path, "vocab.pkl"),'rb'))
args.vocab_size_v1=len(args.vocab)
args.text_features=text_features
args.text_features_1=text_features_1
args.dense_features=dense_features
args.linear_layer_size=[1024,512]
args.text_dim=sum([x[-1] for x in text_features])
args.text_dim_1=sum([x[-1] for x in text_features_1])
args.output_dir="saved_models/index_{}".format(args.index)
#读取word2vector模型
args.embeddings_tables={}
for x in args.text_features:
if x[0] not in args.embeddings_tables:
try:
args.embeddings_tables[x[0]]=gensim.models.KeyedVectors.load_word2vec_format(x[0],binary=False)
except:
args.embeddings_tables[x[0]]=pickle.load(open(x[0],'rb'))
args.embeddings_tables_1={}
for x in args.text_features_1:
if x[0] not in args.embeddings_tables_1:
try:
args.embeddings_tables_1[x[0]]=gensim.models.KeyedVectors.load_word2vec_format(x[0],binary=False)
except:
args.embeddings_tables_1[x[0]]=pickle.load(open(x[0],'rb'))
#读取数据
train_df=pd.read_pickle('data/train_user.pkl')
train_df['label']=train_df['age']*2+train_df['gender']
test_df=pd.read_pickle('data/test_user.pkl')
test_df['label']=test_df['age']*2+test_df['gender']
df=train_df[args.dense_features].append(test_df[args.dense_features])
ss=StandardScaler()
ss.fit(df[args.dense_features])
train_df[args.dense_features]=ss.transform(train_df[args.dense_features])
test_df[args.dense_features]=ss.transform(test_df[args.dense_features])
test_dataset = TextDataset(args,test_df)
#建立模型
skf=StratifiedKFold(n_splits=5,random_state=2020,shuffle=True)
model=ctrNet.ctrNet(args)
#训练模型
for i,(train_index,test_index) in enumerate(skf.split(train_df,train_df['label'])):
if i!=args.index:
continue
logger.info("Index: %s",args.index)
train_dataset = TextDataset(args,train_df.iloc[train_index])
dev_dataset=TextDataset(args,train_df.iloc[test_index])
model.train(train_dataset,dev_dataset)
dev_df=train_df.iloc[test_index]
#输出结果
accs=[]
for f,num in [('age',10),('gender',2)]:
model.reload(f)
if f=="age":
dev_preds=model.infer(dev_dataset)[0]
else:
dev_preds=model.infer(dev_dataset)[1]
for j in range(num):
dev_df['{}_{}'.format(f,j)]=np.round(dev_preds[:,j],4)
acc=model.eval(dev_df[f].values,dev_preds)['eval_acc']
accs.append(acc)
if f=="age":
test_preds=model.infer(test_dataset)[0]
else:
test_preds=model.infer(test_dataset)[1]
logger.info("Test %s %s",f,np.mean(test_preds,0))
logger.info("ACC %s %s",f,round(acc,5))
out_fs=['user_id','age','gender','predict_{}'.format(f)]
out_fs+=['{}_{}'.format(f,i) for i in range(num)]
for i in range(num):
test_df['{}_{}'.format(f,i)]=np.round(test_preds[:,i],4)
test_df['predict_{}'.format(f)]=np.argmax(test_preds,-1)+1
try:
os.system("mkdir submission")
except:
pass
test_df[out_fs].to_csv('submission/submission_test_{}_{}_{}.csv'.format(f,args.index,round(acc,5)),index=False)
logger.info(" best_acc = %s",round(sum(accs),4))