自动拆分并生成列

Question

df

fruit                want_split
apple    colour:red, like:yes, favourite:no
banana   colour:yellow, like:yes, favourite:no
pear     colour:green, like:yes, favourite:yes

有没有办法通过逗号分割然后自动生成列名，而不是手动生成？ 预期输出：

fruit                want_split                  colour      like    favourite
apple    colour:red, like:yes, favourite:no     red        yes       no
banana   colour:yellow, like:yes, favourite:no  yellow     yes       no
pear     colour:green, like:yes, favourite:yes  green      yes       yes

Answer 1

您可以在str.split 和explode 之后使用pivot / unstack 和join：

u = df['want_split'].str.split(",").explode().str.split(":",expand=True)
out = df.join(u.set_index(0,append=True)[1].unstack())

---

print(out)

     type                             want_split  favourite  like  colour
0   apple     colour:red, like:yes, favourite:no         no   yes     red
1  banana  colour:yellow, like:yes, favourite:no         no   yes  yellow
2    pear  colour:green, like:yes, favourite:yes        yes   yes   green

Answer 2

您可以使用.str.extractall 将regex 模式中的捕获组提取为数据框中的列。

c = ['colour','like','favourite']
df[c] = df['want_split'].str.extractall(r':([^,]+)')[0].unstack()

你也可以试试.str.findall:

df[c] = df['want_split'].str.findall(r':([^,]+)').tolist()

---

    fruit                             want_split  colour like favourite
0   apple     colour:red, like:yes, favourite:no     red  yes        no
1  banana  colour:yellow, like:yes, favourite:no  yellow  yes        no
2    pear  colour:green, like:yes, favourite:yes   green  yes       yes

Answer 3

让我们尝试 split 和 dict ，然后创建附加列

out = df.join(pd.DataFrame(df.want_split.map(lambda s : dict(item.split(":") for item in s.split(", "))).tolist(),index=df.index))
    fruit                           want_split  colour like favourite
0   apple     colour:red,like:yes,favourite:no     red  yes        no
1  banana  colour:yellow,like:yes,favourite:no  yellow  yes        no
2    pear  colour:green,like:yes,favourite:yes   green  yes       yes