Skip to content
Laviyx Learn

Try

Python for Data course menu

NumPy Indexing, Slicing and Boolean Masks

Select data from NumPy arrays: indexing and slicing 1D and 2D arrays, rows and columns, boolean masks for filtering, fancy indexing with lists of positions, and the difference between views and copies.

Beginner7 min readUpdated
On this page

Selecting the right part of an array is something you will do constantly. NumPy supports Python-style indexing and slicing, plus two powerful extras: boolean masks and fancy indexing.

1D indexing and slicing#

Works like a Python list: indexes start at 0, negative indexes count from the end, and slices exclude the stop position.

Python
import numpy as np

a = np.array([10, 20, 30, 40, 50, 60])
print(a[0], a[-1])
print(a[1:4])
print(a[::2])
print(a[::-1])
Output
10 60
[20 30 40]
[10 30 50]
[60 50 40 30 20 10]

2D: rows and columns#

Use one pair of brackets with a comma: arr[rows, cols]. A colon on its own means "everything along this dimension".

Python
import numpy as np

m = np.array([
    [ 1,  2,  3,  4],
    [ 5,  6,  7,  8],
    [ 9, 10, 11, 12],
])
print(m[1, 2])          # one element
print(m[0])             # first row
print(m[:, 1])          # second column
print(m[:2, 2:])        # top-right 2 x 2 block
print(m[-1, ::-1])      # last row reversed
Output
7
[1 2 3 4]
[ 2  6 10]
[[3 4]
 [7 8]]
[12 11 10  9]

Boolean masks#

Comparing an array gives a boolean array of the same shape. Using it as an index keeps only the elements where it is True. This is how you filter data in NumPy (and, as you will see, in pandas):

Python
import numpy as np

temps = np.array([18, 25, 31, 22, 35, 29, 16])
hot = temps > 28
print(hot)
print(temps[hot])
print(temps[(temps >= 20) & (temps <= 30)])
print(temps[~hot])          # ~ means "not"
Output
[False False  True False  True  True False]
[31 35 29]
[25 22 29]
[18 25 22 16]

You can also assign through a mask, for example to cap extreme values:

Python
import numpy as np

readings = np.array([12, 250, 18, -5, 30])
readings[readings > 100] = 100
readings[readings < 0] = 0
print(readings)
print(np.clip(np.array([12, 250, 18, -5, 30]), 0, 100))   # the same in one call
Output
[ 12 100  18   0  30]
[ 12 100  18   0  30]

Fancy indexing#

Pass a list (or array) of positions to pick several elements in any order:

Python
import numpy as np

names = np.array(["Asha", "Ravi", "Meera", "John", "Priya"])
scores = np.array([88, 72, 95, 64, 81])

order = np.argsort(scores)[::-1]       # positions from highest score to lowest
print(order)
print(names[order])
print(scores[[0, 2, 4]])
Output
[2 0 4 1 3]
['Meera' 'Asha' 'Priya' 'Ravi' 'John']
[88 95 81]

np.argsort returns the positions that would sort the array; indexing another array with them sorts both together.

Views and copies#

A slice of a NumPy array is a view: it shares memory with the original, so changing it changes the original. Boolean and fancy indexing return copies.

Python
import numpy as np

a = np.array([1, 2, 3, 4, 5])
part = a[1:4]
part[0] = 99
print(a)               # changed through the view

b = np.array([1, 2, 3, 4, 5])
safe = b[1:4].copy()
safe[0] = 99
print(b)               # unchanged
Output
[ 1 99  3  4  5]
[1 2 3 4 5]

Python list slices are copies, so this behaviour surprises many people. Call .copy() when you want an independent array.

Common mistakes#

Practice#

Practice exercise

Daily step counts for two weeks are in a 2 × 7 array: [[4200, 8100, 10500, 3900, 12000, 9800, 6100], [7300, 11000, 5600, 9900, 13400, 2800, 8800]]. Print: (1) the second week, (2) every Wednesday (third column), (3) all days with at least 10,000 steps, and (4) how many such days there were.

Show hint

arr[1], arr[:, 2], a boolean mask arr >= 10000 used as an index (it flattens the result), and .sum() on the mask.

Show solution
Python
import numpy as np

steps = np.array([
    [4200, 8100, 10500, 3900, 12000, 9800, 6100],
    [7300, 11000, 5600, 9900, 13400, 2800, 8800],
])
print(steps[1])
print(steps[:, 2])
print(steps[steps >= 10000])
print((steps >= 10000).sum())
Output
[ 7300 11000  5600  9900 13400  2800  8800]
[10500  5600]
[10500 12000 11000 13400]
4

Key takeaway#

Next: Aggregations and Random Numbers.

Finished this lesson?

Related lessons

Python for DataNumPy Arrays for Beginners: Create Arrays, dtype and shapeLearn NumPy arrays: why they are faster than Python lists, creating arrays with array, zeros, ones, arange and linspace, data types (dtype), shape and ndim, and 2D arrays.Beginner6 min readPython for DataNumPy Vectorisation and Broadcasting ExplainedLearn vectorised operations in NumPy: element-wise arithmetic, comparison operators, universal functions (ufuncs) like sqrt and exp, broadcasting rules for arrays of different shapes, and why vectorised code is faster than loops.Beginner7 min readPython for DataFilter Rows in pandas: Conditions, isin, between and queryFilter pandas DataFrames with boolean conditions, combine them with & | ~, use isin, between and str.contains, write readable filters with query, and count or select the matching rows.Beginner9 min read