在pandas框架中将所有浮动转换为int时遇到问题

Question

chunk_iter = pd.read_csv(source, chunksize=3000)
for chunk in chunk_iter:
    floats = chunk.select_dtypes(include=['floating'], exclude=None)  # select only columns that are float
    notna= floats.columns[floats.notna().any()].tolist()                 # grab the name of all float cols without NaN
    floats[notna] = floats[notna].astype(int)
    print ('converted items in these {} columns to int'.format(notna))

这是我的代码，我认为它应该能够获取所有的浮动列并将它们转换为int但是我收到一个错误，表明那里的某处有一个Nan或inf值。

我想我有点困惑，因为我会认为notna变量会找到所有带有NaN的列并排除它们？