CODEWITHSUNDEEP
pythonarrayspython-3.xnumpyruntime-error
twallien
twallien

Reputation: 141

RuntimeWarning: invalid value encountered in long_scalars

I'm having an issue with modifying an array, by adding the percentage of each item compared to its row to a new matrix. This is the code providing error:

for j in range(1,27):
        for k in range(1,27):
                let_prob[j,k] = let_mat[j,k]*100/(let_mat[j].sum())

I get the error:

RuntimeWarning: invalid value encountered in long_scalars

I have tried rounding the denominator to no success.

Upvotes: 12

Views: 53117

Answers (2)

Gouvernathor
Gouvernathor

Reputation: 147

If it provides any help for other persons encountering this warning, it appeared in a similar context, where I was dividing a value with another, and the warning appeared in cases where I was dividing by 0. Adding a special-case with the value of the denominator, and avoiding a zero-division, made the warning disappear.

I don't know why that warning appeared instead of a ZeroDivisionError.

Upvotes: 8

Kasravnd
Kasravnd

Reputation: 107287

It seems that you are dealing with big numbers, since it raised the error RuntimeWarning. To get rid of such errors, as a numpythonic way you can first calculate the sum of each row using the np.sum() function by specifying the proper axis then repeat and reshape the array in order to be able to divide with your array, them multiple with 100 and round the result:

col, row = np.shape(let_mat) 
let_prob = np.round((let_mat/np.repeat(let_mat.sum(axis=1),row).reshape(col, row).astype(‌​float))*100,2)

Demo :

>>> a = np.arange(20).reshape(4,5)
>>> 
>>> a
array([[ 0,  1,  2,  3,  4],
       [ 5,  6,  7,  8,  9],
       [10, 11, 12, 13, 14],
       [15, 16, 17, 18, 19]])

>>> np.round((a/np.repeat(a.sum(axis=1),5).reshape(4,5).astype(float))*100,2)
array([[  0.  ,  10.  ,  20.  ,  30.  ,  40.  ],
       [ 14.29,  17.14,  20.  ,  22.86,  25.71],
       [ 16.67,  18.33,  20.  ,  21.67,  23.33],
       [ 17.65,  18.82,  20.  ,  21.18,  22.35]])

Upvotes: 3

Related Questions