计算熊猫中相应值的频率[python 3]

Question

I have the dataset with the following values:

var1  var2 
1234  abc
2345  bcs
5678  csd
1234  abc
1234  bcs
5678  csd
1234  bcs
1234  xyz
1234  abc
9101  zzz

I need for every unique value in column var1 to count and show the top 3 frequency counts of the corresponding values in var2, and get the output, for example:

var1  var2  count
1234  abc   3
1234  bcs   2
1234  xyz   1
5678  csd   2
9101  zzz   1

What's the most efficient way of doing that?

Answer 1

You need to include nlargest

df.groupby('var1').var2.apply(lambda x: x.value_counts().nlargest(3)) \
    .reset_index(name='count').rename(columns={'level_1': 'var2'})

   var1 var2  count
0  1234  abc      3
1  1234  bcs      2
2  1234  xyz      1
3  2345  bcs      1
4  5678  csd      2
5  9101  zzz      1

Answer 2

df_a.groupby(['var1','var2'])['var2'].agg({'count':'count'}).reset_index()

Answer 3

这工作：

df.groupby(['var1','var2']).count()