時間序列數據分析--Time Series--時間序列數據統計—滑動窗口

滑動窗口

• 在時間窗口上計算各種統計函數
• 窗口函數
  • 滾動統計    obj.rolling().func
  • 窗口大小 window
  • 窗口是否居中統計  center

##窗口函數--cumsum()
import pandas as pd
import numpy as np

ser_obj = pd.Series(np.random.randn(1000), 
                    index=pd.date_range('20170101', periods=1000))
ser_obj = ser_obj.cumsum()
print(ser_obj.head())
answer
2017-01-01    0.956250
2017-01-02    0.279960
2017-01-03   -0.629871
2017-01-04    0.797499
2017-01-05    0.557393
Freq: D, dtype: float64

r_obj = ser_obj.rolling(window=5)
print(r_obj)
answer
Rolling [window=5,center=False,axis=0]  ####返回的是類對象，需要接統計函數才能展示

print(r_obj.mean().head())
answer：
2017-01-01         NaN
2017-01-02         NaN
2017-01-03         NaN
2017-01-04         NaN
2017-01-05    0.392246
2017-01-06    0.017798
2017-01-07   -0.175184
2017-01-08   -0.137657
2017-01-09   -0.209815
2017-01-10   -0.237701
Freq: D, dtype: float64

# 前5個數據的均值
print(ser_obj[0:5].mean())
answer
0.39224619085269624

# 1-6個數據的均值
print(ser_obj[1:6].mean())
answer：
0.017798268782202475

# 畫圖查看
import matplotlib.pyplot as plt
%matplotlib inline

plt.figure(figsize=(15, 5))

ser_obj.plot(style='r--')
ser_obj.rolling(window=10).mean().plot(style='b')

• 居中顯示，將第5個值，從第三個開始

• print(ser_obj.rolling(window=5, center=True).mean())
  answer：
  2017-01-01          NaN
  2017-01-02          NaN
  2017-01-03     0.392246
  2017-01-04     0.017798
  2017-01-05    -0.175184
  2017-01-06    -0.137657
  2017-01-07    -0.209815
  2017-01-08    -0.237701
  2017-01-09     0.109869
  2017-01-10     0.614796
  2017-01-11     1.203267
  2017-01-12     1.782855