Introduction

Numpy is python library that provides computation on large array elements and matrices. Numpy provides fast and efficient processing on n-dimensional arrays. Array elements in numpy are stored in continuous memory location so that the processing of element is fast and efficeint, unlike in list where the elements are stored in random memory location.

Installation of numpy

Make sure that your system has python and PIP installed already

pip install numpy

 

Creation of numpy array

First of all, we've to import numpy module 

import numpy as np

Here, numpy module is imported as np means numpy is shortened to np.

 

One dimensional array

Let's create a numpy array

arr = np.array([1,3,5,6,8,9])
print("Array Created: ", arr)

Output

Array Created:  [1 3 5 6 8 9]

Now, let's see the type of arr we just created

print("Type of arr is: ", type(arr))

Output

Type of arr is:  <class 'numpy.ndarray'>

 We can create one dimensional array using arange() function

arr = np.arange(10)
print("Array created: ", arr)

Output

Array created:  [0 1 2 3 4 5 6 7 8 9]

arange function in numpy is similar to range() function in python

 

Using list we can create one dimensional array too

a = [1,2,3,4,5]
arr = np.array(a)
print("Array created from list: ", arr)

Output

Array created from list:  [1 2 3 4 5]

 

Two dimensional array

a  = [1,2,3,4,5]
b = [6,7,8,9,10]
arr = np.array([a, b])
print( arr)

Output

[[ 1  2  3  4  5]
 [ 6  7  8  9 10]]

This is how we can create two dimensional array in python using list

arr = np.array([[1,2,3,4,5],[6,7,8,9,10]])
print( arr)

Output

[[ 1  2  3  4  5]
 [ 6  7  8  9 10]]

This is another method where we can give elements directly to create two dimensional array

 

Shape of an array

We can determine the shape of array using shape() function

arr = np.array([[1,2,3,4,5],[6,7,8,9,10]])
print("Shape of array is: ", arr.shape)

Output

Shape of array is:  (2, 5)

Shape function returns the no of rows and columns present in array. In this case there are 2 rows and 5 columns

 

Dimension of an array

arr = np.array([[1,2,3,4,5],[6,7,8,9,10]])
print("Dimension of array is: ", arr.ndim)

Output

Dimension of array is:  2

 

Let's see another examples

arr = np.array([1,2,3,4,5])
print("Dimension of array is: ", arr.ndim)

Output

Dimension of array is:  1

 

arr = np.array([[[1,2,3,4,5], [1,2,3,4,5]], [[1,2,3,4,5], [1,2,3,4,5]]])
print(arr)
print("Dimension of array is: ", arr.ndim)

Output

[[[1 2 3 4 5]
  [1 2 3 4 5]]

 [[1 2 3 4 5]
  [1 2 3 4 5]]]
Dimension of array is:  3

 

Size of an array

a = [1,3,5,6,8,9,6,8]
arr = np.array(a)
print("size of arr is: ", arr.size)

Output

size of arr is:  8

Size function returns the number of elements in an array

 

arr = np.array([[[1,2,3,4,5], [1,2,3,4,5]], [[1,2,3,4,5], [1,2,3,4,5]]])
print("size of array is: ", arr.size)

Output

size of array is:  20

 

Accessing elements of an array

Array elements can be accessed using index same as list and tuple

a = [1,2,3,4,5]
arr = np.array(a)
print(arr[0])
print(arr[2])
print(arr[4])

Output

1
3
5

Elements can also be accessed using loop

for ele in arr:
    print(ele)

Output

1
2
3
4
5

Let's see example of 2-D array

arr = np.array([[1,2,3,4,5], [23, 45, 67 ,98, 100]])
print(arr[0][1])
print(arr[0][0])
print(arr[1][3])
print(arr[1][0])

Output

2
1
98
23
Explanation

In above example, we passed two lists inside single list where first list index as 0 and another one index as 1 to make 2-dimensional array with 2 rows and 5 columns. Thus, if we have to access element from first row two indices must be passed i.e first index for selection of row and second index for selecting column. The first 'print statement' prints the value of first row and second column which is 2. Like as, third 'print statement' prints element from second row and fourth cloumn which is 98.

 

Slicing array

Syntax
array_name[start : end : step]

Array slicing is similar to list slicing in python.

 

Let's see slicing of one dimensional array

a = [1,2,3,4,5]
arr = np.array(a)
print(arr[0:4])

Output

array([1, 2, 3, 4])

Index of array starts with 0 and end with one less than length of an array. Here in this example index starts from 0 and end with 4. We know the end index or upper bound is exclusive the above example retrieve the elements indexing 0 to 3.

 

print(arr[-5:-1])

Output

array([1, 2, 3, 4])

Indexing is assigned in negative as well. The negative indexing starts from -1 which is assigned to last element of array and ends with negative of length of an array.

 

Let's see slicing of two dimensional array
Syntax
array_name[start_row : end_row: step_row, start_column : end_column: step_column]

 

arr = np.array([[1,2,3],[4,5,6],[7,8,9]])
print(arr)

Output

[[1 2 3]
 [4 5 6]
 [7 8 9]]

 

Now, let's take elements from first two rows and last two columns

print(arr[0:2, 1:])

Output

[[2 3]
 [5 6]]

 

Lets take out last element of array

print(arr[2:, 2:])

Output

[[9]]

 

Lets take out last two element from second row

print(arr[1:2, 1:])

Output

[[5 6]]

 

Reshaping an array

Using reshape() function, we can define new array from previously defined array

arr1 = np.array([1,2,3,4,5,6])
arr2 = arr1.reshape(3,2)
print(arr2)

Output

[[1 2]
 [3 4]
 [5 6]]

 

arr1 = np.array([1,2,3,4,5,6])
print("shape of arr1: ", arr1.shape, "\n")
arr2 = arr1.reshape(3,2)
print(arr2, "\n")
print("shape of arr2: ", arr2.shape)

Output

shape of arr1:  (6,) 

[[1 2]
 [3 4]
 [5 6]] 

shape of arr2:  (3, 2)

Here we've changed shape of arr1. All we have to care about during reshaping is that the no of elements must be same in both new and previous array

 

arr2 = arr1.reshape(2,2)
print(arr2)

Output

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-70-a75024147a88> in <module>
      1 arr1 = np.array([1,2,3,4,5,6])
----> 2 arr2 = arr1.reshape(2,2)
      3 print(arr2)

ValueError: cannot reshape array of size 6 into shape (2,2)

Here, we got ValueError because we are trying to reshaping array having 6 element to array with 4 element in it.

 

arr1 = np.array([[1,2,3,4,5,6], [7,8,9,10,11,12]])
arr2 = arr1.reshape(12,1)
print(arr2)

Output

[[ 1]
 [ 2]
 [ 3]
 [ 4]
 [ 5]
 [ 6]
 [ 7]
 [ 8]
 [ 9]
 [10]
 [11]
 [12]]

In both arrays, the number of element are same just shape is changed.

 

arr2 = arr1.reshape(4, 3)
print(arr2)

Output

[[ 1  2  3]
 [ 4  5  6]
 [ 7  8  9]
 [10 11 12]]

 

Appending & Inserting row and column in array

Using append() function

 

Row wise appending
Syntax
np.append(previous_array, [array_to_be_add], axis =0)

 

a = np.array([20,21,22])
np.append(arr2,[a],axis=0)

Output

array([[ 1,  2,  3],
       [ 4,  5,  6],
       [ 7,  8,  9],
       [10, 11, 12],
       [20, 21, 22]])

 

Column wise appending
Syntax
 np.append(previous_array, array_to_be_add, axis =1)

 

a = np.array([20,21,22,23])
b= a.reshape(4,-1)
np.append(arr2,b,axis=1)

Output

array([[ 1,  2,  3, 20],
       [ 4,  5,  6, 21],
       [ 7,  8,  9, 22],
       [10, 11, 12, 23]])

 

Using insert()function

 

Row wise inserting
Syntax
np.insert(previous_array, inserting_index, array_tobe_inserted, axis=0)

 

arr = np.array([[1,2,3],[4,5,6],[7,8,9]])
inserting_arr = np.array([11,12,13])
print("Before insertion: ")
print(arr)
print("after insertion at index 2: ")
print(np.insert(arr, 2, inserting_arr, axis=0))

Output

Before insertion: 
[[1 2 3]
 [4 5 6]
 [7 8 9]]
after insertion at index 2: 
[[ 1  2  3]
 [ 4  5  6]
 [11 12 13]
 [ 7  8  9]]

 

Column wise inserting
Syntax
np.insert(previous_array, inserting_index, array_tobe_inserted, axis=1)

 

arr = np.array([[1,2,3],[4,5,6],[7,8,9]])
inserting_arr = np.array([11,12,13])
print("Before insertion: ")
print(arr)
print("after insertion at index 1: ")
print(np.insert(arr, 1, inserting_arr, axis=1))

Output

Before insertion: 
[[1 2 3]
 [4 5 6]
 [7 8 9]]
after insertion at index 1: 
[[ 1 11  2  3]
 [ 4 12  5  6]
 [ 7 13  8  9]]

 

Matrix generation using numpy

We can generate martices having elements all one and zero using ones() and zeros() function

zero_matrix = np.zeros([3,3], dtype=int)
print(zero_matrix)

Output

[[0 0 0]
 [0 0 0]
 [0 0 0]]

 

Let's take another example

ones_matrix = np.ones([4,5], dtype=float)
print(ones_matrix)       

Output

[[1. 1. 1. 1. 1.]
 [1. 1. 1. 1. 1.]
 [1. 1. 1. 1. 1.]
 [1. 1. 1. 1. 1.]]

 

Using random, we can generate metrices too

Syntax
np.random.rand(size of array)

 

arr = np.random.rand(3,5)
print(arr)
print(arr.shape)

Output

[[0.74206847 0.44733595 0.10237527 0.34372174 0.87838503]
 [0.48042584 0.46966427 0.318181   0.88341896 0.46838867]
 [0.02591508 0.58777176 0.07273747 0.80669176 0.69172011]]
(3, 5)

 

Element wise operation of array

One dimensional array
arr1 = np.array([1,2,3,4])
arr2 = np.array([4,3,2,1])
print("sum: ", arr1+arr2)
print("Difference: ", arr1-arr2)
print("Multiplication: ", arr1*arr2)
print("Division: ", arr1/arr2)

Output

sum:  [5 5 5 5]
Difference:  [-3 -1  1  3]
Multiplication:  [4 6 6 4]
Division:  [0.25       0.66666667 1.5        4.        ]

 

Two dimensional array
arr1 = np.array([[1,2,3,4], [4,3,2,1]])
arr2 = np.array([[4,3,2,1], [4,3,2,1]])
print("sum:\n ", arr1+arr2)
print("Difference:\n ", arr1-arr2)
print("Multiplication:\n ", arr1*arr2)
print("Division:\n ", arr1/arr2)

Output

sum:
  [[5 5 5 5]
 [8 6 4 2]]
Difference:
  [[-3 -1  1  3]
 [ 0  0  0  0]]
Multiplication:
  [[ 4  6  6  4]
 [16  9  4  1]]
Division:
  [[0.25       0.66666667 1.5        4.        ]
 [1.         1.         1.         1.        ]]

 

Conclusion

Numpy is a powerful library that provides fast computation on large array of element and matrices. Elements are stored in continuous memory location so the processing of array element is faster than in list. Numpy has usecase in image processing too as opencv sees images as array of 1's and 0's. Numpy can be used with matplotlib library to plot various bar charts, histogram etc. So numpy is the powerful and most useable library in python.

Happy Learning :-)

 

Reference

https://numpy.org/