1 df.groupby.count 

different with

df.groupby.iloc[a_col].count


2 nan become nat when concat to a df which the col conrespond type is time