Introduction
Numpy is a 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 contiguous memory locations so that the processing of elements is fast and efficient, unlike in a list where the elements are stored in a 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 the numpy module
import numpy as np
Here, the 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]
Note: arange() function in numpy is similar to range() function in python
Now, let’s use a list to create an array using numpy
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
Previously we’ve seen how a one-dimensional array is created using the numpy module in python. Now, we’re going to see how a two-dimensional array is created using numpy
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 a two-dimensional array in python using a 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
The shape of an array
We can determine the shape of an 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)
The shape function returns the no of rows and columns present in an array. In this case, there are 2 rows and 5 columns
Dimension of an array
we can find the dimension of a numpy 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 example
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
we can also check the size of a numpy 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
The 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 a loop
for ele in arr:
print(ele)
Output
1 2 3 4 5
Let’s see an example of a 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 the above example, we passed two lists inside a single list where the first list index is 0 and another one index is 1 to make a 2-dimensional array with 2 rows and 5 columns.
Thus, if we have to access elements from the first row two indices must be passed i.e. the first index for the selection of row and the 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 elements from the second row and fourth column which is 98.
Slicing array
Syntax
array_name[start : end : step]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 retrieves 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 a negative 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]]
Let’s take out last two elements from the second row
print(arr[1:2, 1:])
Output
[[5 6]]
Reshaping an array
Using reshape() function, we can define a new array from a 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 the shape of arr1. All we have to care about during reshaping is that the no of elements must be the same in both the 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 reshape an array having 6 elements to an array with 4 elements 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 elements is same just the 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 rows and columns in the 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 matrices having elements all one and zero using the 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 matrices 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 an 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 a large array of elements and matrices. Elements are stored in contiguous memory locations so the processing of array elements is faster than in the list.
Numpy has a use case in image processing too as OpenCV sees images as an array of 1’s and 0’s. Numpy can be used with the matplotlib library to plot various bar charts, histograms, etc. So numpy is the powerful and most useable library in python.
Happy Learning 🙂
Reference
