为什么这些 Pandas groupby 上的差异会导致 Nan？

Question

The example dataframe I have is-

>>> new_df
        date country  score
0 2018-01-01      ch     50
1 2018-01-01      es    100
2 2018-01-01      us    150
3 2018-01-02      ch     10
4 2018-01-02      gb    100
5 2018-01-02      us    125
6 2018-01-03      us    160

Why does new_df.groupby(["date", "country"]).diff() produce Nan?

>>> new_df.groupby(["date", "country"]).diff()
   score
0    NaN
1    NaN
2    NaN
3    NaN
4    NaN
5    NaN
6    NaN

Answer 1

As you can see the size of each group is 1 , then the subnetting of the subtraction is NaN because to make the subtraction a minuend and a subtraend are needed, that is to say size at least equal to 2 :

df.groupby(['date','country']).size()

date        country
2018-01-01  ch         1
            es         1
            us         1
2018-01-02  ch         1
            gb         1
            us         1
2018-01-03  us         1
dtype: int64

Answer 2

It's because there is nothing to subtract--you have only one value per group in your example.