Here I have a simplified dataframe (The real one is in the same format but just amplified)

import pandas as pd
import numpy as np

row = (1, 2)
columns = ["x", "y", "x", "y", "x", "y", "x", "y"]
data = ([10, 2, 8, 1.5, 9, 2, 11, 1.6], [8, 3, 7.5, 2.2, 9, 2, 8.6, 2.3])

df = pd.DataFrame(data, index = row, columns = columns)

I want to calculate the variance of x, y for both of 1, 2, and the ideal format is

Any hint or help is appreciated

CodePudding user response：

Try this -

1. Unstack to get the x, y columns as indexes
2. groupby over both the levels [x,y] and [1,2] and calculate variance.
3. Unstack and transpose to get [x,y] as columns.

df.unstack().groupby(level=[0,1]).var().unstack().T

          x         y
1  1.666667  0.069167
2  0.435833  0.189167