假设我有一个系列和数据帧,如:
import pandas as pd
s = pd.Series([10,20,11,12,30,34],
index=["red","red","blue","blue","green","green"])
s.index.name="numbers"
df = pd.DataFrame({
"color":["red","green","blue","blue","red","green"],
"id":[1,2,3,4,5,6]})
我想将s
中的值添加到df
中的列中,顺序与它们出现的顺序相同,其中s
的索引等于df["color"]
,即
pd.some_function(df,s,left_on="color",right_index=True)
color id numbers
red 1 10
green 2 30
blue 3 11
blue 4 12
red 5 20
green 6 34
我尝试过pd.merge
、pd.join
等,但我根本无法使其工作(如果不在df
上循环,由color
过滤,添加来自s
的数据,然后在最后将其连接(
您可以使用groupby.cumcount
为merge
:设置唯一密钥
idx1 = s.groupby(level=0).cumcount()
# [0, 1, 0, 1, 0, 1]
idx2 = df.groupby('color').cumcount()
# [0, 0, 0, 1, 1, 1]
s.index.name="color"
out = (df
.merge(s.reset_index(name='number'),
left_on=['color', idx2], right_on=['color', idx1])
.drop(columns='key_1')
)
变体:
s.index.name="color"
out = (df
.assign(idx=df.groupby('color').cumcount())
.merge(s.reset_index(name='number')
.assign(idx=s.groupby(level=0).cumcount().values),
left_on=['color', 'idx'], right_on=['color', 'idx'])
.drop(columns='idx')
)
输出:
color id number
0 red 1 10
1 green 2 30
2 blue 3 11
3 blue 4 12
4 red 5 20
5 green 6 34