I think I would prefer to use an isinstance check: In [11]: df.loc[df.A.apply(lambda d: isinstance(d, dict))] Out[11]: A 2 {'a': 5, 'b': 6, ...
確定! 回上一頁