python - 定制 Pandas 宽到长

Question

我有这种类型的宽数据集，我想将其转换为长格式。 choice 列显示了所选选项的顺序。因此，如果它是 choice2 ，则表示选择了第二个选项，并且该列中的第二个选项应为 1，其他选项应为 0。 college, hsg2 coml5 列应重复。

数据:

respondent  choice  college hsg2    coml5   type1   type2   type3   type4   type5   type6   fuel1   fuel2   fuel3   fuel4   fuel5   fuel6
1   choice1 0   0   0   van regcar  van stwagon van truck   cng cng electric    electric    gasoline    gasoline
2   choice2 1   1   1   regcar  van regcar  stwagon regcar  truck   methanol    methanol    cng cng gasoline    gasoline

这就是我想要实现的目标。我找不到任何方法来正确转换 choice 列。任何帮助表示赞赏。我查看了大部分从宽到长，反之亦然的问题，但无法解决我的问题。

Answer 1

使用wide_to_long首先，然后从 choice 列中获取数字，并按最后一列的每组计数器进行比较:

df = pd.wide_to_long(df, 
                     stubnames=['type','fuel'], 
                     i=['respondent','choice','college','hsg2','coml5'],
                     j='tmp').reset_index().drop('tmp', 1)

df['choice'] = df['choice'].str.extract('(\d+)').astype(int)

df['order'] = df.groupby(['respondent','choice','college','hsg2','coml5']).cumcount().add(1)

df['new'] = df['choice'].eq(df['order']).astype(int)
print (df)
    respondent  choice  college  hsg2  coml5     type      fuel  order  new
0            1       1        0     0      0      van       cng      1    1
1            1       1        0     0      0   regcar       cng      2    0
2            1       1        0     0      0      van  electric      3    0
3            1       1        0     0      0  stwagon  electric      4    0
4            1       1        0     0      0      van  gasoline      5    0
5            1       1        0     0      0    truck  gasoline      6    0
6            2       2        1     1      1   regcar  methanol      1    0
7            2       2        1     1      1      van  methanol      2    1
8            2       2        1     1      1   regcar       cng      3    0
9            2       2        1     1      1  stwagon       cng      4    0
10           2       2        1     1      1   regcar  gasoline      5    0
11           2       2        1     1      1    truck  gasoline      6    0

如果order是根据type和fuel列名称生成的:

df = pd.wide_to_long(df, 
                     stubnames=['type','fuel'], 
                     i=['respondent','choice','college','hsg2','coml5'],
                     j='order').reset_index()

df['choice'] = df['choice'].str.extract('(\d+)').astype(int)
df['new'] = df['choice'].eq(df['order']).astype(int)
print (df)
    respondent  choice  college  hsg2  coml5  order     type      fuel  new
0            1       1        0     0      0      1      van       cng    1
1            1       1        0     0      0      2   regcar       cng    0
2            1       1        0     0      0      3      van  electric    0
3            1       1        0     0      0      4  stwagon  electric    0
4            1       1        0     0      0      5      van  gasoline    0
5            1       1        0     0      0      6    truck  gasoline    0
6            2       2        1     1      1      1   regcar  methanol    0
7            2       2        1     1      1      2      van  methanol    1
8            2       2        1     1      1      3   regcar       cng    0
9            2       2        1     1      1      4  stwagon       cng    0
10           2       2        1     1      1      5   regcar  gasoline    0
11           2       2        1     1      1      6    truck  gasoline    0