一、滑動平均

公式（窗口為7）：

也就是說7個數做一次平均

二、指數平均

公式：

其中

因此

也就是說離本點越近，考慮的權重也越大。

python code：

# author: adrian.wu
import numpy as np
from matplotlib import pyplot as plt
import pandas as pd

"""
count : count(1)
day: d
"""
path = ‘/Users/adrian.wu/PycharmProjects/jobs/time_series/data/ord_count_by_day.csv
 
‘

data_raw = pd.read_csv(path)
data_raw.sort_values("d", inplace=True)

"""
draw raw time series picture
"""
# plt.plot(data_raw[‘d‘], data_raw[‘count(1)‘])
# plt.show()

win_7 = data_raw[‘count(1)‘].rolling(window=7).mean()
ewma_30 = data_raw[‘count(1)‘].ewm(span=30).mean()
d = [i for i in
 
 range(0, len(ewma_30))]

fig, ax = plt.subplots(1, 1)

ax.plot(d, data_raw[‘count(1)‘], label=‘raw_data‘)
ax.plot(d, win_7, label=‘win_7‘)
ax.plot(d, ewma_30, label=‘ewma_span=30‘)
plt.legend()
plt.show()

時間序列預測（一）數據的一些處理方法