'bitwise_and' not supported for the input types Numpy array

Question

I am trying to calculate F1_Score using numpy array using the code below

ypred = np.squeeze(imgs_mask_predict[jj,:,:,:])
ytrue = np.squeeze(imgs_test_mask[jj,:,:,:])

def f1_score_single(y_true, y_pred):
    y_true = y_true.flatten('F')
    y_pred = y_pred.flatten('F')
    cross_size = len(y_true & y_pred)
    if cross_size == 0: return 0.
    p = 1. * cross_size / len(y_pred)
    r = 1. * cross_size / len(y_true)
    return (2. * (p * r) / (p + r))

I get an error

"ufunc 'bitwise_and' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe'' " at this line of code

    cross_size = len(y_true & y_pred)

I tried to search for this error but did not get the reason and solution for this. How should I resolve this?

y_true type <class 'numpy.ndarray'> y_pred type <class 'numpy.ndarray'> y_true shape (256, 160) y_pred shape (256, 160) — Kanvas
– Kanvas, Commented Feb 14, 2018 at 14:20
That looks like type(y_true) and y_true.shape. Now show us y_true.dtype and y_pred.dtype. The dtype attribute tells you the data type of the data in the numpy array. — Warren Weckesser
– Warren Weckesser, Commented Feb 14, 2018 at 14:21
Ah. Why are you trying to use the & operator (which for numpy arrays is bitwise-and) with floating point values? That operator only makes sense (for some definition of "sense") with integer values. — Warren Weckesser
– Warren Weckesser, Commented Feb 14, 2018 at 14:25

relay · Accepted Answer · 2018-02-14 22:51:11Z

2

You should use isclose for floating point numbers.

cross_size = np.isclose(y_true, y_pred).sum()

You can also set a threshold for similarity by atol and rtol keyword arguments, you can see the documentation here.

answered Feb 14, 2018 at 22:51

relay

1998 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Shaina Raza · Accepted Answer · 2020-05-29 18:36:06Z

0

you can try intersect1d instead of &, for example, the use of np.intersect1d will do the same as you intend to do with &

import numpy as np
y_true=np.array([1.0,2.0,4.0,1.0,2.0,4.0,1.0,2.0,4.0,1.0])
y_score=np.array([1.0,1.0,2.0,4.0,1.0,2.0,4.0,1.0,2.0,4.0])
x=np.intersect1d(y_true, y_score)
print(x)

the result will be [1. 2. 4.]

you can test the program here https://www.programiz.com/python-programming/online-compiler/

answered May 29, 2020 at 18:36

Shaina Raza

1,67820 silver badges13 bronze badges

Collectives™ on Stack Overflow

'bitwise_and' not supported for the input types Numpy array

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related