stock correlation analysis

stock_corr.py

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import tushare as ts

#获取数据
s_pf = '600000'
s_gd = '601818'
sdate = '2016-01-01'
edate = '2016-12-31'
df_pf = ts.get_h_data(s_pf, start = sdate, end = edate).sort_index(axis = 0, ascending = True)
df_gd = ts.get_h_data(s_gd, start = sdate, end = edate).sort_index(axis = 0, ascending = True)
df = pd.concat([df_pf.close, df_gd.close], axis = 1, keys = ['pf_close','gd_close'])
df.ffill(axis = 0, inplace = True) #填充数据

df.to_csv('pf_gd.csv')


# 分析数据
df = pd.read_csv('pf_gd.csv')

corr = df.corr(method = 'pearson', min_periods = 1)
print(corr)

df.plot(figsize = (20,12))
plt.savefig('pf_gd.png')
plt.close()

df['pf_one'] = df.pf_close / float(df.pf_close[0]) * 100
df['gd_one'] = df.gd_close / float(df.gd_close[0]) * 100
df.pf_one.plot(figsize = (20,12), label='pf')
df.gd_one.plot(figsize = (20,12), label='gd')
plt.legend(shadow=True, fancybox=True)
plt.savefig('pf_gd_one.png')
plt.close()