python - Grouping and finding most frequent values -

i have df this:

protein peptide        aaa        aaa        aba b        aaa b        aba b        aba 

but need filter data finding each value in column 1 top occurring value in column 2.

so output like:

protein peptide        aaa b        aba 

in reality need top 3 occuring values. don't know how solve using python , pandas?

mode isn't groupby method, though series (and dataframe) method, have pass apply:

in [11]: df.groupby('protein')['peptide'].apply(lambda x: x.mode()[0]) out[11]: protein    aaa b    aba name: peptide, dtype: object 

---

to top three, use value_counts (in same way):

in [12]: df.groupby('protein')['peptide'].apply(lambda x: x.value_counts()[:3]) out[12]: protein        aaa    2          aba    1 b        aba    2          aaa    1 dtype: int64