Boolean Indexing

Numpy has the option to select elements with conditions. 

Use the following syntax: ary[ary operator number]

• ary[ary > 5]
• ary[ary == 5]
• ary[ary <= 5]
• etc

Example 1

Retrieve all numbers under 5

x = np.array([1, 5, 2, 4, 9 ,10 ,13, 7])
x[x < 5]
>>> [1, 2, 4]

Example 2

What happens with higher dimensions?

x = np.array([[1, 5], [2, 4], [9 ,10] ,[13, 7]])
x[x < 5]  # --> [1, 2, 4]
x[x >= 5] # --> [5, 9, 10, 13, 7]
x[x == 6] # --> []

Did you see that? It returned a single array with all the elements that satisfy the condition

When we apply a condition on x it returns a boolean array that represents for each index if it satisfies the condition.

Example 3

This time we will examine what happens if we simply use a condition without the brackets []

x = np.array([1, 2, 3])
x < 2
>>> [True, False, False]

It returns a list of all the indices that satisfy the condition

Example 4

Instead of placing a condition inside the brackets [], We can instead place a boolean list just like we received in the previous example to retrieve the desired elements.

x = np.array([8, 4, 10])
x[[False, True, False]]  # --> [4]
x[[True, True, False]]   # --> [8, 4]
x[[True, False, True]]   # --> [8, 10]
x[[False, False, False]] # --> []

Example 3 and 4 explain why x[x < 5] works.