python - 在 python 中将 Series reshape 为 Dataframe 矩阵

Question

我有一个包含 305 个条目的系列，这些条目带有数据时间索引。数据看起来像这样

1992-01-31     1.123077 
1992-02-28    -2.174845 
1992-03-31    -3.884848
1992-04-30     8.682919
1992-05-29     1.312976
1992-06-30     7.851080
1992-07-31    -3.192788
1992-08-31    -7.351976
1992-09-30    -6.782217
1992-10-30   -17.182738
1992-11-30     3.898782
1992-12-31   -26.190414
1993-01-29     2.233359
1993-02-26     6.709006
continues with monthly data till December 2017

我想将数据 reshape 为一个 DataFrame，其中包含行的所有年份和列的所有月份以及要根据需要填写的数据

        January February  March     etc >>  December
2017    values  values    values    values  values  
2016    values  values    values    values  values  
2015    values  values    values    values  values  
etc \\// 
1992    values                  

我查看了其他帖子并尝试了 reshape 和 asmatrix，但鉴于它是不均匀的系列，我不断收到此错误。

ValueError:新数组的总大小必须不变。

我真正想做的是，如果矩阵是奇数形状，则为缺失值插入 NaN。因此，如果 2017 年没有 11 月或 12 月的值，它们将为 NaN

如果有人能帮上忙，请告诉我

Answer 1

来源 DF:

In [159]: df
Out[159]:
                  val
date
1992-01-31   1.123077
1992-02-28  -2.174845
1992-03-31  -3.884848
1992-04-30   8.682919
1992-05-29   1.312976
1992-06-30   7.851080
1992-07-31  -3.192788
1992-08-31  -7.351976
1992-09-30  -6.782217
1992-10-30 -17.182738
1992-11-30   3.898782
1992-12-31 -26.190414
1993-01-29   2.233359
1993-02-26   6.709006

解决方法:

import calendar

In [158]: (df.assign(year=df.index.year, mon=df.index.month)
             .pivot(index='year', columns='mon', values='val')
             .rename(columns=dict(zip(range(13), calendar.month_name))))
Out[158]:
mon    January  February     March     April       May     June      July    August  September    October  November   December
year
1992  1.123077 -2.174845 -3.884848  8.682919  1.312976  7.85108 -3.192788 -7.351976  -6.782217 -17.182738  3.898782 -26.190414
1993  2.233359  6.709006       NaN       NaN       NaN      NaN       NaN       NaN        NaN        NaN       NaN        NaN

更新: 或者更好更短的 version from @COLDSPEED :

In [164]: pd.pivot(df.index.year, df.index.month, df['val']) \
            .rename(columns=calendar.month_name.__getitem__)
Out[164]:
date   January  February     March     April       May     June      July    August  September    October  November   December
date
1992  1.123077 -2.174845 -3.884848  8.682919  1.312976  7.85108 -3.192788 -7.351976  -6.782217 -17.182738  3.898782 -26.190414
1993  2.233359  6.709006       NaN       NaN       NaN      NaN       NaN       NaN        NaN        NaN       NaN        NaN