我有一個包含多列的大熊貓數(shù)據(jù)框，但是讓我們關注兩個：df = pd.DataFrame([['hey how are you', 'fine thanks',1], ['good to know', 'yes, and you',2], ['I am fine','ok',3], ['see you','bye!',4]],columns=list('ABC'))df出去： A B C0 hey how are you fine thanks 11 good to know yes, and you 22 I am fine ok 33 see you bye! 4從前面的數(shù)據(jù)框中，我如何將兩個特定的列壓縮成一個單獨的 Pandas 數(shù)據(jù)框來執(zhí)行其他列的值？例如： A C0 hey how are you 11 fine thanks 12 good to know 23 yes, and you 24 I am fine 35 ok 36 see you 47 bye! 4我試過了：df = df['A'].stack()df = df.groupby(level=0)df但是，它不起作用。知道如何實現(xiàn)新格式嗎？
3 回答
慕后森
TA貢獻1802條經(jīng)驗 獲得超5個贊
這將刪除列名，但完成工作：
import pandas as pd
df = pd.DataFrame([['hey how are you', 'fine thanks'],
['good to know', 'yes, and you'],
['I am fine','ok'],
['see you','bye!']],columns=list('AB'))
df.stack().reset_index(drop=True)
0 hey how are you
1 fine thanks
2 good to know
3 yes, and you
4 I am fine
5 ok
6 see you
7 bye!
dtype: object
默認堆棧行為保留列名：
df.stack()
0 A hey how are you
B fine thanks
1 A good to know
B yes, and you
2 A I am fine
B ok
3 A see you
B bye!
dtype: object
如果您有更多列，您可以選擇要堆疊的列，只需使用列索引：
df[["A", "B"]].stack()
使用額外的列，事情變得棘手，您需要通過降低一級（包含列）來對齊索引：
df["C"] = range(4)
stacked = df[["A", "B"]].stack()
stacked.index = stacked.index.droplevel(level=1)
stacked
0 hey how are you
0 fine thanks
1 good to know
1 yes, and you
2 I am fine
2 ok
3 see you
3 bye!
dtype: object
現(xiàn)在我們可以連接C列：
pd.concat([stacked, df["C"]], axis=1)
0 C
0 hey how are you 0
0 fine thanks 0
1 good to know 1
1 yes, and you 1
2 I am fine 2
2 ok 2
3 see you 3
3 bye! 3
白衣染霜花
TA貢獻1796條經(jīng)驗 獲得超10個贊
您可能正在尋找的是pandas.concat.
它接受“Series、DataFrame 或 Panel 對象的序列或映射”，因此您可以傳遞選擇列list的DataFrame對象（將pd.Series在為單個列編制索引時）。
df3 = pd.concat([df['A'], df['B']])
添加回答
舉報
0/150
提交
取消