How to change array (or matrix) type with numpy in python ?


Examples of how to change array (or matrix) type with numpy in python

Change the type of an existing matrix

Let's consider the following matrix of integer type:

import numpy as np

A = np.array([[10, 20, 30], [60, 20, 10], [50, 30, 90]])

print(A)
print(A.dtype)

donne ici

[[10 20 30]
[60 20 10]
[50 30 90]]

et

int64

To change the type, a solution is to use astype (see numpy.ndarray.dtype)

A = A.astype('float64')

print(A)
print(A.dtype)

returns

[[10. 20. 30.]
[60. 20. 10.]
[50. 30. 90.]]

and

float64

Initialization of a matrix with a given type

It is also possible to specify the type of a matrix during the creation:

import numpy as np

A = np.array([[1, 2, 3]], dtype=float)

print(A)

print(A.dtype)

returns

[[1. 2. 3.]]

and

float64

Combine matrix with different type

It is important to check the type of a matrix to avoid loosing information, an example let's consider the following matrix A:

A = np.array([[10, 20, 30], [60, 20, 10], [50, 30, 90]])

returns

[[10 20 30]
 [60 20 10]
 [50 30 90]]

and the matrix B:

B= np.array([[2.1, 7.3, 4.5]])

returns

[[2.1 7.3 4.5]]

Now, if the matrix A is updated using B, like:

A[1,:] = B

returns

[[10 20 30]
 [ 2  7  4]
 [50 30 90]]

but the elements of B have been modified. To avoid that a solution would have been to do first:

A = A.astype('float64')

A[1,:] = B

returns

[[10.  20.  30. ]
 [ 2.1  7.3  4.5]
 [50.  30.  90. ]]

References