(2026.04.04 Sat @崇左)

滑動窗口計(jì)算 pandas.rolling

pandas.rolling方法用于計(jì)算DataFrame中Series或DataFrame對象的滑動窗口的各項(xiàng)指標(biāo)，聚合和變換函數(shù)包括min, max, sum, mean，對時(shí)間序列分析和平滑噪聲數(shù)據(jù)尤其有效。

案例：

import pandas as pd
import numpy as np

df = pd.DataFrame({'value': [10, 20, 30, np.nan, 50, 60]})

# 計(jì)算窗口尺寸為3的均值
rolling_mean = df['value'].rolling(window=3, min_periods=1).mean()
print(rolling_mean)

0 10.000000
1 15.000000
2 20.000000
3 25.000000
4 40.000000
5 55.000000
Name: value, dtype: float64

可自定義參數(shù)：

• window： 窗口尺寸
• min_periods： 用于計(jì)算值的最小觀察區(qū)間，默認(rèn)值和window值相同
• center：Align labels at the center of the window if True.
• win_type：應(yīng)用加權(quán)窗，e.g., triang, guassian，默認(rèn)均勻加權(quán)
• on：用于基于時(shí)間的窗的字段
• step (pandas 1.5+)：Evaluate every nth step in the rolling window.

返回的結(jié)果和原Series的尺寸相同，無法賦值的部分用NaN填充。

pandas.resample

pandas.resample/df.resample方法用于對時(shí)間序列進(jìn)行重采樣，允許用戶將一個(gè)時(shí)間序列從一種頻率轉(zhuǎn)換為另一種頻率，并可對常規(guī)時(shí)間序列進(jìn)行聚合或計(jì)算統(tǒng)計(jì)值。

案例：

import pandas as pd

# 創(chuàng)建
dates = pd.date_range(start='2023-01-01', end='2023-12-31', freq='D')
data = np.random.rand(len(dates))
df = pd.DataFrame(data, index=dates, columns=['Random Data'])

# 重采樣改為月度數(shù)據(jù)
monthly_resampled_data = df.resample('M').mean()
print(monthly_resampled_data.head())

參數(shù)：

• rule： The offset string or object representing target conversion.
• axis： Which axis to use for up- or down-sampling.
• closed: Which side of bin interval is closed.
• label: Which bin edge label to label bucket with.
• convention: For PeriodIndex only, controls whether to use the start or end of rule.
• on: For a DataFrame, column to use instead of index for resampling. Column must be datetime-like.
• level: For a MultiIndex, level (name or number) to use for resampling. Level must be datetime-like.