A numpy array with 0, 1 elements (also called mask array) can be used to mask some elements in a new array. This trick is very useful when we are operating numpy array.
In this tutorial, we will introduce how to convert and create a numpy mask array based on a given array.
numpy.where()
We will use numpy.where()
It is defined as:
numpy.where(condition, [x, y, ]/)
When condition is True, it will return x, otherwise, return y.
Create a msk array based on threshold value
if x and y is scalar
Example 1：
import numpy as np
x1 = np.array(range(16))
x1 = x1.reshape([4,4])
print("x1=",x1)
y1 = np.where(x1<=4, 1, 0)
print(y1)
It outputs:
x1= [[ 0 1 2 3] [ 4 5 6 7] [ 8 9 10 11] [12 13 14 15]] y1= [[1 1 1 1] [1 0 0 0] [0 0 0 0] [0 0 0 0]]
In this example, all elements that are bigger than 4 will be assigned to 0, others are assigned to 1.
Example 2:
y2 = np.where(np.logical_and(x>=4, x<=8), 1, 0)
print("y2=", y2)
It outputs:
y2= [[0 0 0 0] [1 1 1 1] [1 0 0 0] [0 0 0 0]]
In this exampe, elemens in x1 are in [4, 8] will be assigned to 1.
If x and y are numpy array.
import numpy as np
x1 = np.array(range(16))
x1 = x1.reshape([4,4])
print("x1=",x1)
x_mask = x1*3
y_mask = x1*-3
y1 = np.where(x1<=4, x_mask, y_mask)
print("y1=",y1)
y2 = np.where(np.logical_and(x1>=4, x1<=8), x_mask, y_mask)
print("y2=", y2)
It outputs:
x1= [[ 0 1 2 3] [ 4 5 6 7] [ 8 9 10 11] [12 13 14 15]] y1= [[ 0 3 6 9] [ 12 -15 -18 -21] [-24 -27 -30 -33] [-36 -39 -42 -45]] y2= [[ 0 -3 -6 -9] [ 12 15 18 21] [ 24 -27 -30 -33] [-36 -39 -42 -45]]
In this exampe, the shapes of x_mask and y_mask are same to x1.
You can use numpy.where() to get y1 and y2 mask array based on different condition.
Moreover, if you do not use numpy.where(), you can read:
Convert Boolean to 0 and 1 in NumPy