Merge pull request #1939 from hutuxian/din_bug_fix

update README and fix print_function

fluid/PaddleRec/din/README.md

@@ -36,6 +36,10 @@ DIN通过一个兴趣激活模块(Activation Unit)，用预估目标Candidate AD
 ```
 cd data && sh data_process.sh && cd ..
 ```
+如果执行过程中遇到找不到某个包（例如pandas包）的报错，使用如下命令安装对应的包即可。
+```
+pip install pandas
+```
 
 * Step 2: 产生训练集、测试集和config文件
 ```
@@ -103,6 +107,12 @@ model saved in din_amazon/global_step_50000
 ...
 ```
 
+提示：
+
+* 在单机条件下，使用代码中默认的超参数运行时，产生最优auc的global step大致在440000到500000之间
+
+* 训练超出一定的epoch后会稍稍出现过拟合
+
 ## 预测
 参考如下命令，开始预测.
 

fluid/PaddleRec/din/data/build_dataset.py

+from __future__ import print_function
 import random
 import pickle
-from __future__ import print_function
 
 random.seed(1234)
 

fluid/PaddleRec/din/data/convert_pd.py

+from __future__ import print_function
 import pickle
 import pandas as pd
 
@@ -13,10 +14,12 @@ def to_df(file_path):
         return df
 
 
+print("start to analyse reviews_Electronics_5.json")
 reviews_df = to_df('./raw_data/reviews_Electronics_5.json')
 with open('./raw_data/reviews.pkl', 'wb') as f:
     pickle.dump(reviews_df, f, pickle.HIGHEST_PROTOCOL)
 
+print("start to analyse meta_Electronics.json")
 meta_df = to_df('./raw_data/meta_Electronics.json')
 meta_df = meta_df[meta_df['asin'].isin(reviews_df['asin'].unique())]
 meta_df = meta_df.reset_index(drop=True)

fluid/PaddleRec/din/data/remap_id.py

+from __future__ import print_function
 import random
 import pickle
 import numpy as np
-from __future__ import print_function
 
 random.seed(1234)
 

fluid/PaddleRec/din/train.py

@@ -121,7 +121,7 @@ def train():
                 loss_sum = 0.0
 
                 if (global_step > 400000 and global_step % PRINT_STEP == 0) or (
-                        global_step < 400000 and global_step % 50000 == 0):
+                        global_step <= 400000 and global_step % 50000 == 0):
                     save_dir = args.model_dir + "/global_step_" + str(
                         global_step)
                     feed_var_name = [