为什么 Pandas 的DataFrame.apply方法在DataFrame为空时调用正在应用的函数?
例如:
>>> import pandas as pd
>>> df = pd.DataFrame({"foo": []})
>>> df
Empty DataFrame
Columns: [foo]
Index: []
>>> x = []
>>> df.apply(x.append, axis=1)
Series([], dtype: float64)
>>> x
[Series([], dtype: float64)] # <<< why was the apply callback called with an empty row?
Run Code Online (Sandbox Code Playgroud)
深入研究 Pandas 源代码,看起来这就是罪魁祸首:
if not all(self.shape):
# How to determine this better?
is_reduction = False
try:
is_reduction = not isinstance(f(_EMPTY_SERIES), Series)
except Exception:
pass
if is_reduction:
return Series(NA, index=self._get_agg_axis(axis))
else:
return self.copy()
Run Code Online (Sandbox Code Playgroud)
看起来 Pandas 正在调用不带参数的函数,试图猜测结果应该是 aSeries还是 a DataFrame。
我想补丁已经准备好了。
编辑:此问题已被修补,现在已记录并允许reduce使用选项来避免它:http://pandas.pydata.org/pandas-docs/dev/ generated/pandas.DataFrame.apply.html