这一节的前面部分的操作内容及LSTM的堆叠都与上一节完全一样，只是我们最终用途上有差别。03讲的是classification，04讲的是regression。
导入各类包
import warnings
warnings.filterwarnings('ignore')
%matplotlib inline
from pathlib import Path
import numpy as np
import pandas as pd
from scipy.stats import spearmanr
import tensorflow as tf
from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping
from tensorflow.keras.models import Sequential, Model
from tensorflow.keras.layers import Dense, LSTM, Input, concatenate, Embedding, Reshape, BatchNormalization
import tensorflow.keras.backend as K
import matplotlib.pyplot as plt
from matplotlib.ticker import FuncFormatter
import seaborn as sns
gpu_devices = tf.config.experimental.list_physical_devices('GPU')
if gpu_devices:
print('Using GPU')
tf.config.experimental.set_memory_growth(gpu_devices[0], True)
else:
print('Using CPU')
idx = pd.IndexSlice
sns.set_style('whitegrid')
np.random.seed(42)#到这一步都是之前已经做过的操作进行一个预处理。
results_path = Path('results', 'lstm_embeddings')#从路径导入新数据。
if not results_path.exists():
results_path.mkdir(parents=True)导入数据
这里导入的数据就是我们在01中所创建的数据集。
data = pd.read_hdf('data.h5', 'returns_weekly').drop('label', axis=1)#删除表中的label行。
data['ticker'] = pd.factorize(data.index.get_level_values('ticker'))[0]#依旧对刻度标准进行分类。
data['month'] = data.index.get_level_values('date').month#index是索引，get_level_values是返回所请求级别的值索引，也就是按照date的索引进行一个返回。
data = pd.get_dummies(data, columns=['month'], prefix='month')#对原数据集进行一个添加拓展。
data.info()<class 'pandas.core.frame.DataFrame'>
MultiIndex: 1547552 entries, ('000001', Timestamp('2017-12-31 00:00:00')) to ('689009', Timestamp('2008-01-02 00:00:00'))
Data columns (total 66 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 0 1547552 non-null float64
1 1 1547552 non-null float64
2 2 1547552 non-null float64
3 3 1547552 non-null float64
4 4 1547552 non-null float64
5 5 1547552 non-null float64
6 6 1547552 non-null float64
7 7 1547552 non-null float64
8 8 1547552 non-null float64
9 9 1547552 non-null float64
10 10 1547552 non-null float64
11 11 1547552 non-null float64
12 12 1547552 non-null float64
13 13 1547552 non-null float64
14 14 1547552 non-null float64
15 15 1547552 non-null float64
16 16 1547552 non-null float64
17 17 1547552 non-null float64
18 18 1547552 non-null float64
19 19 1547552 non-null float64
20 20 1547552 non-null float64
21 21 1547552 non-null float64
22 22 1547552 non-null float64
23 23 1547552 non-null float64
24 24 1547552 non-null float64
25 25 1547552 non-null float64
26 26 1547552 non-null float64
27 27 1547552 non-null float64
28 28 1547552 non-null float64
29 29 1547552 non-null float64
30 30 1547552 non-null float64
31 31 1547552 non-null float64
32 32 1547552 non-null float64
33 33 1547552 non-null float64
34 34 1547552 non-null float64
35 35 1547552 non-null float64
36 36 1547552 non-null float64
37 37 1547552 non-null float64
38 38 1547552 non-null float64
39 39 1547552 non-null float64
40 40 1547552 non-null float64
41 41 1547552 non-null float64
42 42 1547552 non-null float64
43 43 1547552 non-null float64
44 44 1547552 non-null float64
45 45 1547552 non-null float64
46 46 1547552 non-null float64
47 47 1547552 non-null float64
48 48 1547552 non-null float64
49 49 1547552 non-null float64
50 50 1547552 non-null float64
51 51 1547552 non-null float64
52 52 1547552 non-null float64
53 label 1547552 non-null int64
54 ticker 1547552 non-null int64
55 month_1 1547552 non-null uint8
56 month_2 1547552 non-null uint8
57 month_3 1547552 non-null uint8
58 month_4 1547552 non-null uint8
59 month_5 1547552 non-null uint8
60 month_6 1547552 non-null uint8
61 month_7 1547552 non-null uint8
62 month_8 1547