from scipy.stats import variation 

A = np.random.randn(10, 10)

# max variation along rows of A; 
# rows: axis=0, cols: axis=1

var = variation(A, axis=0)
idmax = np.argmax(var)
print(idmax)