pandas aggregate count in dataframe

Question

I have a DataFrame and I am using .aggregate({'col1': np.sum}), this will perform a summation of the values in col1 and aggregate them together. Is it possible to perform a count, something like .aggregate({'col1': some count function here})?

`{'col1': 'count'}` or `{'col1': 'size'}` or `{'col1': 'nunique'}` depending on your use case. — root, Jan 16 '17 at 17:53
Or `len` (the built-in), which I suggest is the most readable of the bunch. — Aleksey Bilogur, Jan 16 '17 at 17:57
`len` is typically slower than `'size'`, as it's a python built-in instead of numpy under the hood. — root, Jan 16 '17 at 18:06

root · Accepted Answer · 2017-01-16T18:08:58.747

You can use 'size', 'count', or 'nunique' depending on your use case. The differences between them being:

'size': the count including NaN and repeat values.
'count': the count excluding NaN but including repeats.
'nunique': the count of unique values, excluding repeats and NaN.

For example, consider the following DataFrame:

df = pd.DataFrame({'col0': list('aabbcc'), 'col1': [1, 1, 2, np.nan, 3, 4]})

  col0  col1
0    a   1.0
1    a   1.0
2    b   2.0
3    b   NaN
4    c   3.0
5    c   4.0

Then using the three functions described:

df.groupby('col0')['col1'].agg(['size', 'count', 'nunique'])

      size  count  nunique
col0                      
a        2      2        1
b        2      1        1
c        2      2        2

pandas aggregate count in dataframe

1 Answers1