将一个系列分配给 Pandas DataFrame 的几行 [英] Assign a Series to several Rows of a Pandas DataFrame

问题描述

我有一个用索引和列准备的 Pandas DataFrame，所有值都是 NaN.现在我计算了一个结果，它可以用于一个以上的 DataFrame 行，我想一次分配它们.这可以通过一个循环来完成，但我很确定这个任务可以一次完成.

I have a pandas DataFrame prepared with an Index and columns, all values are NaN. Now I computed a result, which can be used for more than one row of a DataFrame, and I would like to assign them all at once. This can be done by a loop, but I am pretty sure that this assignment can be done at once.

这是一个场景:

import pandas as pd
df = pd.DataFrame(index=['A', 'B', 'C'], columns=['C1', 'C2'])  # original df
s = pd.Series({'C1': 1, 'C2': 'ham'})  # a computed result
index = pd.Index(['A', 'C'])  # result is valid for rows 'A' and 'C'

幼稚的方法是

df.loc[index, :] = s

但这根本不会改变DataFrame.它仍然是

But this does not change the DataFrame at all. It remains as

    C1   C2
A  NaN  NaN
B  NaN  NaN
C  NaN  NaN

如何完成这项任务?

推荐答案

看来可以用底层数组数据来赋值了-

It seems we can use the underlying array data to assign -

df.loc[index, :] = s.values

现在，这里假设 s 中的索引顺序与 df 的列中的索引顺序相同.如果不是这种情况，如 @Nras 建议，我们可以使用 s[df.columns].values 进行右侧赋值.

Now, this assumes that the order of index in s is same as in the columns of df. If that's not the case, as suggested by @Nras, we could use s[df.columns].values for right side assignment.

这篇关于将一个系列分配给 Pandas DataFrame 的几行的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！