如何组合 pandas pivot 表中的列？

Question

I have a pivot table with 3 levels of columns. For each unique mean and std, I want to combine them into a str f"{x.mean}({x.std})" replacing the mean and std columns with the new mean_std_str column.

Heres a print of the dataframe:

rescore_func   asp                 chemscore  ... goldscore   plp                
tag           best      first           best  ...     first  best      first     
              mean  std  mean  std      mean  ...       std  mean  std  mean  std
dock_func                                     ...                                
asp           65.2  0.7  34.5  2.4      64.0  ...       0.0  64.4  0.7  37.9  0.7
chemscore     59.1  2.0  29.5  2.0      58.0  ...       1.7  58.7  0.7  40.9  2.3
goldscore     68.9  1.7  34.8  4.3      69.7  ...       1.3  68.9  1.3  46.2  0.7
plp           69.3  1.1  35.2  2.0      69.7  ...       2.0  68.9  2.4  39.4  2.9

[4 rows x 16 columns]

Desired output:

rescore_func   asp                             
tag            best       first           ...  
               mean_std   mean_std        ...
dock_func                                 ...                                 
asp            65.2(0.7)  34.5(2.4)       ...
chemscore      59.1(2.0)  29.5(2.0)       ... 
goldscore      68.9(1.7)  34.8(4.3)       ...
plp            69.3(1.1)  35.2(2.0)       ...

[4 rows x 16 columns]

So far I have:

df = df.melt(ignore_index=False).reset_index()
df = df.rename(columns=str).rename(columns={'None':'descr'})

which gives:

    dock_func rescore_func    tag descr  value
0         asp          asp   best  mean   65.2
1   chemscore          asp   best  mean   59.1
2   goldscore          asp   best  mean   68.9
3         plp          asp   best  mean   69.3
4         asp          asp   best   std    0.7
..        ...          ...    ...   ...    ...
59        plp          plp  first  mean   39.4
60        asp          plp  first   std    0.7
61  chemscore          plp  first   std    2.3
62  goldscore          plp  first   std    0.7
63        plp          plp  first   std    2.9

[64 rows x 5 columns]

I am stuck on how to group the means and std together before re-pivoting the data...

Answer 1

DataFrame.reorder_levels will make it easy for you.

Here is some sample data:

import numpy as np
import pandas as pd


index = pd.Index(["asp", "chemscore", "goldscore", "plp"], name="dock_func")
columns = pd.MultiIndex.from_product(
    [index, pd.Index(["best", "fisrt"], name="tag"), ("mean", "std")]
)

df = pd.DataFrame(
    np.random.random(size=(4, 16)),
    index=index,
    columns=columns,
).round(1)

df looks like:

dock_func  asp                 chemscore                 goldscore                  plp
tag       best      fisrt           best      fisrt           best      fisrt      best      fisrt
          mean  std  mean  std      mean  std  mean  std      mean  std  mean  std mean  std  mean  std
dock_func
asp        0.5  0.6   0.4  0.2       0.7  0.7   0.8  0.1       0.2  0.5   0.6  0.7  0.5  0.2   0.2  0.7
chemscore  0.0  0.7   0.9  0.2       0.3  0.3   0.4  0.8       0.3  0.4   0.2  0.8  0.5  0.5   0.4  0.2
goldscore  0.5  0.7   0.8  0.0       0.2  0.8   0.1  0.2       0.6  0.1   0.4  0.2  0.8  0.2   0.8  0.3
plp        1.0  0.6   0.6  0.8       0.8  0.6   0.3  1.0       0.7  0.2   0.8  0.2  0.2  0.2   0.7  0.2

Then just run the following:

df = df.reorder_levels([2, 0, 1], axis=1).astype(str)
df = df["mean"] + "(" + df["std"] + ")"

and df is:

dock_func       asp           chemscore           goldscore                 plp
tag            best     fisrt      best     fisrt      best     fisrt      best     fisrt
dock_func
asp        0.5(0.6)  0.4(0.2)  0.7(0.7)  0.8(0.1)  0.2(0.5)  0.6(0.7)  0.5(0.2)  0.2(0.7)
chemscore  0.0(0.7)  0.9(0.2)  0.3(0.3)  0.4(0.8)  0.3(0.4)  0.2(0.8)  0.5(0.5)  0.4(0.2)
goldscore  0.5(0.7)  0.8(0.0)  0.2(0.8)  0.1(0.2)  0.6(0.1)  0.4(0.2)  0.8(0.2)  0.8(0.3)
plp        1.0(0.6)  0.6(0.8)  0.8(0.6)  0.3(1.0)  0.7(0.2)  0.8(0.2)  0.2(0.2)  0.7(0.2)