numpy find the max value in a row and return back to it's column index

Question

I have a 180295* 10 numpy array called lda_trans, row means words, and column means 10 topics.

array([[0.01841009, 0.01840699, 0.35798764, ..., 0.38443892, 0.01841072,
        0.12870054],
       [0.1       , 0.1       , 0.1       , ..., 0.1       , 0.1       ,
        0.1       ],
       [0.1       , 0.1       , 0.1       , ..., 0.1       , 0.1       ,
        0.1       ],
       ...,
       [0.0416964 , 0.62473603, 0.0416964 , ..., 0.04169395, 0.04169796,
        0.04169232],
       [0.03772096, 0.03775132, 0.66048403, ..., 0.03771698, 0.03772411,
        0.0377139 ],
       [0.03754747, 0.03756587, 0.66206395, ..., 0.03754399, 0.037551  ,
        0.03753927]])

Now I want to turn back each row's maximum value's column name, I only know how to extract the max value in each row, but I don't know how to get the column name. I know in pandas can use idxmax. But is there any similar function in Numpy? Thanks!

for i in range(180295):
    lda_trans_max.append(np.max(lda_trans[i]))

Answer 1

Use np.argmax.

Demo:

>>> a                                                                                                                            
array([[0, 1, 2, 3, 4],
       [5, 6, 7, 8, 9]])
>>> np.argmax(a, axis=1)                                                                                                             
array([4, 4])

You are getting [4, 4] here because in both rows, the element with the maximum value is at position 4.

Another demo:

>>> a                                                                                                                                
 array([[5, 9, 7, 6, 8],
       [8, 7, 7, 6, 9]])
>>> np.argmax(a, axis=1)                                                                                                             
array([1, 4])