如何连续订购3列？巨蟒熊猫

Question

我有以下数据帧：

df = {names:["CARL","HERMES","LEO"], srv:[234,123,44], mx_l1:[10,12,8], mx_l2:[20,8,10], mx_l3:[12,23,11], label:["street","city","apto"]}

但是长度列(mx_l1、mx_l2、mx_l3)必须垂直排列，所以我保证它们与当前列的顺序。

我如何连续地对这些列进行排序，而不是放在单个列中，如下所示

name  |srv|max_lenght|label
CARL  |234|10        |street
CARL  |234|20        |city
CARL  |234|12        |apto
HERMES|123|12        |street
HERMES|234|8         |city
HERMES|234|23        |apto

我的代码

df2 = df.melt(id_vars=["srv", "name"],
              value_vars=["max_l1", "max_l2", "max_l3"],
              value_name="max_lenght")\
    .sort_values(by="srv")\
    .drop("variable", axis=1)\ 
    .reset_index(drop=True)

Answer 1

IIUC，这是你想要的：

df = (
    df.melt(
        id_vars=["names", "srv", "label"],
        value_vars=["mx_l1", "mx_l2", "mx_l3"],
        value_name="max_length",
    )
    .sort_values(["names", "srv", "max_length"])
    .drop(columns="variable")
    .reset_index(drop=True)
)

print(df)
names  srv   label  max_length
    0    CARL  234  street          10
    1    CARL  234  street          12
    2    CARL  234  street          20
    3  HERMES  123    city           8
    4  HERMES  123    city          12
    5  HERMES  123    city          23
    6     LEO   44    apto           8
    7     LEO   44    apto          10
    8     LEO   44    apto          11

Answer 2

使用set_index和stack

df_final = (df.set_index(['names','srv','label']).stack().droplevel(-1)
              .rename_axis(index={'label':'max_length'})
              .reset_index(name='label'))

Out[31]:
    names  srv max_length  label
0    CARL  234     street     10
1    CARL  234     street     20
2    CARL  234     street     12
3  HERMES  123       city     12
4  HERMES  123       city      8
5  HERMES  123       city     23
6     LEO   44       apto      8
7     LEO   44       apto     10
8     LEO   44       apto     11