我有一个数据框,看起来像这样。
它有8列和n行。第一列是缺少日期的日期。 (例如1946-01-04等),但是也有重复项(例如1946-01-02),我想要一个代码来识别重复项,但还要填写缺失的日期并将NaN
添加到其他代码中单元格中的行。
我尝试过
dfx = pd.DataFrame(None, index=pd.DatetimeIndex(start=df.地震の発生日時.min(), end=df.地震の発生日時.max(), freq='D'))
df = df.apply(pd.concat([df, dfx], join='outer', axis=1))
但是它只是从.min()
到.max()
添加到了文件的末尾...我想将其应用于数据之内,例如
Date Time Places w x y z
1946-01-02 14:45:00 6.8 36.3 140.1 31 3.2 1
1946-01-02 22:18:00 7.6 40.5 141.4 0 4.6 3
1946-01-02 23:29:00 6.7 36.1 139.4 39 4.3 2
1946-01-03 04:28:00 5.6 34.4 136.5 1 4.2 2
1946-01-03 04:36:00 6.5 35.5 139.5 50 3 1
1946-01-04 00:00:00 NaN NaN NaN NaN NaN NaN
1946-01-06 10:56:00 8.1 41.5 143.4 51 5.2 3
顺便说一句。我无法使用inner join
。它抛出:
AttributeError: 'Places' is not a valid function for 'Series' object
答案 0 :(得分:1)
如果第一栏没有被时间填充DatetimeIndex
的解决方案:
print (df)
Time Places w x y z col
Date
1946-01-02 14:45:00 6.8 36.3 140.1 31 3.2 1
1946-01-02 22:18:00 7.6 40.5 141.4 0 4.6 3
1946-01-02 23:29:00 6.7 36.1 139.4 39 4.3 2
1946-01-03 04:28:00 5.6 34.4 136.5 1 4.2 2
1946-01-05 04:36:00 6.5 35.5 139.5 50 3.0 1
print (df.index)
DatetimeIndex(['1946-01-02', '1946-01-02', '1946-01-02', '1946-01-03',
'1946-01-05'],
dtype='datetime64[ns]', name='Date', freq=None)
使用date_range
创建新的DataFrame:
dfx = pd.DataFrame(index=pd.date_range(start=df.index.min(),
end=df.index.max(), freq='D'))
print (dfx)
Empty DataFrame
Columns: []
Index: [1946-01-02 00:00:00, 1946-01-03 00:00:00, 1946-01-04 00:00:00, 1946-01-05 00:00:00]
然后使用DataFrame.join
:
df = dfx.join(df)
print (df)
Time Places w x y z col
1946-01-02 14:45:00 6.8 36.3 140.1 31.0 3.2 1.0
1946-01-02 22:18:00 7.6 40.5 141.4 0.0 4.6 3.0
1946-01-02 23:29:00 6.7 36.1 139.4 39.0 4.3 2.0
1946-01-03 04:28:00 5.6 34.4 136.5 1.0 4.2 2.0
1946-01-04 NaN NaN NaN NaN NaN NaN NaN
1946-01-05 04:36:00 6.5 35.5 139.5 50.0 3.0 1.0
如果有DatetimeIndex
次,请按DataFrame.reset_index
创建一列:
print (df)
Places w x y z col
DateTime
1946-01-02 14:45:00 6.8 36.3 140.1 31 3.2 1
1946-01-02 22:18:00 7.6 40.5 141.4 0 4.6 3
1946-01-02 23:29:00 6.7 36.1 139.4 39 4.3 2
1946-01-03 04:28:00 5.6 34.4 136.5 1 4.2 2
1946-01-05 04:36:00 6.5 35.5 139.5 50 3.0 1
print (df.index)
DatetimeIndex(['1946-01-02 14:45:00', '1946-01-02 22:18:00',
'1946-01-02 23:29:00', '1946-01-03 04:28:00',
'1946-01-05 04:36:00'],
dtype='datetime64[ns]', name='DateTime', freq=None)
df = df.reset_index()
print (df)
DateTime Places w x y z col
0 1946-01-02 14:45:00 6.8 36.3 140.1 31 3.2 1
1 1946-01-02 22:18:00 7.6 40.5 141.4 0 4.6 3
2 1946-01-02 23:29:00 6.7 36.1 139.4 39 4.3 2
3 1946-01-03 04:28:00 5.6 34.4 136.5 1 4.2 2
4 1946-01-05 04:36:00 6.5 35.5 139.5 50 3.0 1
然后用DateTime
列中的misisng值替换Series.str.normalize
,最后merge
删除时间:
d = df['DateTime'].dt.normalize()
dfx = pd.DataFrame({'Dates':pd.date_range(start=d.min(),
end=d.max(), freq='D')})
print (dfx)
Dates
0 1946-01-02
1 1946-01-03
2 1946-01-04
3 1946-01-05
df = dfx.merge(df.assign(Dates=d), on='Dates', how='left')
df['DateTime'] = df['DateTime'].fillna(df['Dates'])
print (df)
Dates DateTime Places w x y z col
0 1946-01-02 1946-01-02 14:45:00 6.8 36.3 140.1 31.0 3.2 1.0
1 1946-01-02 1946-01-02 22:18:00 7.6 40.5 141.4 0.0 4.6 3.0
2 1946-01-02 1946-01-02 23:29:00 6.7 36.1 139.4 39.0 4.3 2.0
3 1946-01-03 1946-01-03 04:28:00 5.6 34.4 136.5 1.0 4.2 2.0
4 1946-01-04 1946-01-04 00:00:00 NaN NaN NaN NaN NaN NaN
5 1946-01-05 1946-01-05 04:36:00 6.5 35.5 139.5 50.0 3.0 1.0