Generating a dataframe from a groupby transformation

March 22, 2022

I have this code:

df.droupby('type)['feature1].mean()

df has 15 features from 1 to 15

I want to iterate over a list of the features names and append these 15 results in a dataframe:

Desired Output: type(dataframe)

    type      feature1       features2  ..... feature15
    type_A    mean(float)    mean(float)      mean(float)
    type_B    mean(float)    mean(float)      mean(float)
    type_c    mean(float)    mean(float)      mean(float)

What I did:

I have the list of features:

list = df.iloc[:, 10:24].columns.to_list()

and tried something like this:

for i in len(list):
    df.groupby('type')[list[i]].mean()

to see if I get something and this line returns an error:

'int' object is not iterable

Can anyone help me with this?

>Solution :

IIUC, you could simply use groupby + mean:

out = df.groupby('type', as_index=False).mean()

But if you have a bunch of other columns that you don’t want to include in the calculation and only want the mean of "feature…" columns, you could filter, then groupby + mean:

out = df.filter(like='feature').groupby(df['type']).mean().reset_index()

Output:

  type  feature1  feature2
0    A      10.0      11.0
1    B      12.0      14.0
2    C      13.0      13.0
3    D      12.0      19.0
4    E      10.0      10.0