我正在處理一個時間序列，該序列記錄了 2013 年至 2021 年巴西城市市場上魚的價格，原始資料集有三列，一列是最便宜的，另一列是最貴的，最后是他們收集資料當天發現的平均價格。我已經為相應的列制作了三個子集，日期并索引了日期，然后進行了一些解釋性分析，我發現 2013 年和 2014 年的某些特定月份具有 nan 值。
dfmin.loc['2013-4-1':'2013-7-31']
min
date
2013-04-01 12:00:00 16.0
2013-04-02 12:00:00 16.0
2013-05-22 12:00:00 NaN
2013-05-23 12:00:00 NaN
2013-05-24 12:00:00 NaN
2013-05-27 12:00:00 NaN
2013-05-28 12:00:00 NaN
2013-05-29 12:00:00 NaN
2013-05-30 12:00:00 NaN
2013-05-31 12:00:00 NaN
2013-06-03 12:00:00 NaN
2013-06-04 12:00:00 NaN
2013-06-05 12:00:00 NaN
2013-06-06 12:00:00 NaN
2013-06-07 12:00:00 NaN
2013-06-10 12:00:00 NaN
2013-06-11 12:00:00 NaN
2013-06-12 12:00:00 NaN
2013-06-13 12:00:00 NaN
2013-06-14 12:00:00 NaN
2013-06-17 12:00:00 NaN
2013-06-18 12:00:00 NaN
2013-06-19 12:00:00 15.8
2013-06-20 12:00:00 15.8
2013-06-21 12:00:00 15.8
?```
I want to fill these NaN values from the month 05 with the average value from the medium price from the month 04 and the month 06, how can I make it?
uj5u.com熱心網友回復：
IIUC，您可以使用簡單的索引：
# if needed, convert to datetime
#df.index = pd.to_datetime(df.index)
df.loc[df.index.month==5, 'min'] = df.loc[df.index.month.isin([4,6]), 'min'].mean()
或者如果您第 5 個月沒有 NaN：
mask = df.index.month==5
df.loc[mask, 'min'] = (df.loc[mask, 'min']
.fillna(df.loc[df.index.month.isin([4,6]), 'min'].mean())
)
輸出：
min
date
2013-04-01 12:00:00 16.00
2013-04-02 12:00:00 16.00
2013-05-22 12:00:00 15.88
2013-05-23 12:00:00 15.88
2013-05-24 12:00:00 15.88
2013-05-27 12:00:00 15.88
2013-05-28 12:00:00 15.88
2013-05-29 12:00:00 15.88
2013-05-30 12:00:00 15.88
2013-05-31 12:00:00 15.88
2013-06-03 12:00:00 NaN
2013-06-04 12:00:00 NaN
2013-06-05 12:00:00 NaN
2013-06-06 12:00:00 NaN
2013-06-07 12:00:00 NaN
2013-06-10 12:00:00 NaN
2013-06-11 12:00:00 NaN
2013-06-12 12:00:00 NaN
2013-06-13 12:00:00 NaN
2013-06-14 12:00:00 NaN
2013-06-17 12:00:00 NaN
2013-06-18 12:00:00 NaN
2013-06-19 12:00:00 15.80
2013-06-20 12:00:00 15.80
2013-06-21 12:00:00 15.80
轉載請註明出處，本文鏈接：https://www.uj5u.com/gongcheng/446264.html