Let's say I have a numpy df such as
X = pd.DataFrame(np.array([
[ 9., 12., 4.],
[ 1., 31., 3.],
[ 6., 4., 1.]]) )
I want to divide each entry by the row sum * column sum. I know I can get row and column sum, respectively, using X.sum() and X.sum(1) but I don't know how to do the operation I'm seeking aside from looping through each entry brute force.
Use broadcasting:
import pandas as pd
import numpy as np
X = pd.DataFrame(np.array([
[9., 12., 4.],
[1., 31., 3.],
[6., 4., 1.]]))
X_arr = X.values
mult = X_arr.sum(1)[:, None] * X_arr.sum(0)
result = X / mult
print(result)
Output
0 1 2
0 0.022500 0.010213 0.020000
1 0.001786 0.018845 0.010714
2 0.034091 0.007737 0.011364
You just need to divide twice:
# axis='rows' divide row-wise
x.div(x.sum()).div(x.sum(1), axis='rows')
Output:
0 1 2
0 0.022500 0.010213 0.020000
1 0.001786 0.018845 0.010714
2 0.034091 0.007737 0.011364
