减少对数据帧压缩列表的迭代

Question

考虑以下代码，该代码使用functools.reduce连接数据帧列表：

from functools import reduce
df1 = pd.DataFrame({'A': [1, 2]})
df2 = pd.DataFrame({'B': [3, 4]})
df3 = pd.DataFrame({'C': [5, 6]})
reduce(lambda x, y: pd.concat([x, y], axis=1), [df1, df2, df3])

此代码运行良好。然而，当我尝试以下操作时，我会出现错误：

reduce(lambda x, y: pd.concat([x[0], y[0]], axis=1), zip([df1, df2, df3], [0, 1, 0]))

有人能帮我理解一下吗？

Answer 1

让我们了解reduce:中发生了什么

# Iteration 1: 
# x = (df1, 0); y = (df2, 1)
# reduce(x, y): pd.concat([x[0], y[0]], axis=1) # okay
# Now the result of `reduce(x, y)` is a dataframe which will be used as new x for iteration 2
# Iteration 2: 
# x = some_dataframe, y = (df3, 0)
# reduce(x, y): pd.concat([x[0], y[0]], axis=1) # error
# Notice that x is not a tuple anymore but a dataframe instead.
# So calling dataframe[0] will raise an key error because there is no such column in the dataframe

如果您对reduce的实现感兴趣，这里是最小的实现：

def reduce(func, sequence):
if not sequence:
raise TypeError('Empty sequence')
result = sequence[0]
for item in sequence[1:]:
result = func(result, item)

return result