高效方法:dfs[dfs['delta'].isnull()==False].sort_values(by='delta', ascending=True).groupby('Call_Number', as_index=False).first()