notebook.community

Edit and run



In [1]:

    
import numpy as np



In [2]:

    
A = np.array([
    [1., 1., 0.25],
    [1., 1., 0.2],
    [4., 5., 1.]
])
AR = np.array([
    [1., 2.],
    [0.5, 1.]
])
AM = np.array([
    [1., 0.5],
    [2., 1.]
])
AA = np.array([
    [1., 1.],
    [1., 1.]
])



In [3]:

    
def norm(X):
    return np.sum(X, axis=1) / np.sum(X)

Let's find a wegith vector for our criterias.



In [4]:

    
W = norm(A)
W









    Out[4]:





array([ 0.15570934,  0.15224913,  0.69204152])

Now we can build a norms for our criterias.



In [5]:

    
X = np.vstack((norm(AR), norm(AM), norm(AA)))
X









    Out[5]:





array([[ 0.66666667,  0.33333333],
       [ 0.33333333,  0.66666667],
       [ 0.5       ,  0.5       ]])

And scale them properly.



In [6]:

    
usefulness = np.dot(X.T, W)
usefulness









    Out[6]:





array([ 0.5005767,  0.4994233])

So that we choose the first one.