df.groupby('column1').column2.nunique()
#column name ISSDTM pd.to_datetime(df.ISSDTM, errors='coerce').dt.year #result 0 2013 1 2013 2 2009 3 2009 Name: ISSDTM, dtype: int64