Numpy is a python library that offers Linear Algebra functionalities. It is a very important library in the Python machine learning stack. This is because all the high-level machine learning and data analysis python libraries use Numpy as their building block. Numpy does all linear algebra operation using a data structure known as Numpy Array.

Numpy Array : numpy.array()

Numpy uses n-dimensional array objects called ndarray a.k.a numpy array, to carry out linear algebra computations.

Numpy array can be formed by passing Python Tuples or Lists to array() function of numpy. This function returns an object of ndarray.

Advantage of Numpy Array

For any computation between python lists, you will have to loop through their elements to perform the operation. This is costly in terms of performance. But numpy array carries out vector operations which is much faster. Due to this great performance, numpy array is used in machine learning algorithm implementation in Python.

Let us see some practical examples of the Numpy Array.

Ad

Deep Learning Specialization on Coursera

Importing Numpy Library

import numpy as np

Example 1: Creating a 1-D Numpy Array from Python List

#Create a Python List and assign it to a variable
py_list = [1,2,3]

#Create the Numpy Array using list variable
numpy_1D_array = np.array(py_list)

numpy_1D_array

array([1, 2, 3])

#Create Numpy Array without python list variable
numpy_1D_array = np.array([1,2,3])
numpy_1D_array

array([1, 2, 3])

Example 2 – Creating 2-D Numpy Array from Python List

#Create a python list containging two list
py_list_of_lists = [[1,2,3,4],[5,6,7,8]] 

#Create the Numpy Array using list variable
numpy_2D_array = np.array(py_list_of_lists)
numpy_2D_array

array([[1, 2, 3, 4],
       [5, 6, 7, 8]])

#Create Numpy Array without python list variable
numpy_2D_array = np.array([[1,2,3,4],[5,6,7,8]])
numpy_2D_array

array([[1, 2, 3, 4],
       [5, 6, 7, 8]])

Example 3 – Creating 3-D Numpy Array from Python List

numpy_3D_array = np.array([[[1,2,3,4],[5,6,7,8]] , [[1,2,3,4],[5,6,7,8]], [[1,2,3,4],[5,6,7,8]]])
numpy_3D_array

array([[[1, 2, 3, 4],
        [5, 6, 7, 8]],

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

       [[1, 2, 3, 4],
        [5, 6, 7, 8]]])

Example 4 – Creating 1-D Numpy Array from Python Tuple

#Create a Python Tuple and assign it to a variable
py_tuple = (1,2,3)

#Create the Numpy Array using list variable
numpy_array = np.array(py_tuple)

numpy_array

array([1, 2, 3])

Data Type : ndarray.dtype

The data type of the numpy array can be determined by using dtype method of ndarray object. Lets us see the two examples below.

numpy_array = np.array([[1, 2, 4], [5, 8, 7]])
numpy_array.dtype

dtype('int32')

numpy_array = np.array([[1.5, 2.7, 4.8], [5.3, 8.1, 7.0]])
numpy_array.dtype

dtype('float64')

Creating Numpy Array with specific Data Type

At times, you would like to explicitly specify the data type of the numpy array. In the first example below, since all elements are integers, the data type on numpy array is int32 by default.

But we can explicitly pass the data type as argument dtype in array() function. This can be seen in the 2nd example.

#Example 1
numpy_array = np.array([[1,2,3,4], [5,6,7,8]])

print("The data type is-", numpy_array.dtype)
numpy_array

The data type is- int32

array([[1, 2, 3, 4],
       [5, 6, 7, 8]])

#Example 2
numpy_array = np.array([[1,2,3,4], [5,6,7,8]], dtype = 'float64')

print("The data type is-", numpy_array.dtype)
numpy_array

The data type is- float64

array([[1., 2., 3., 4.],
       [5., 6., 7., 8.]])

Numpy Array Size : ndarray.size

The size of the numpy array can be determined by using the size method of the ndarray object. ndarray. size returns the number of elements inside the numpy array.

Lets us see the examples below.

numpy_1D_array = np.array([1,2,3])
numpy_1D_array.size

3

numpy_2D_array = np.array([[1,2,3,4],[5,6,7,8]])
numpy_2D_array.size

8

numpy_3D_array = np.array([[[1,2,3,4],[5,6,7,8]] , [[1,2,3,4],[5,6,7,8]], [[1,2,3,4],[5,6,7,8]]])
numpy_3D_array.size

24

Numpy Array is Homogeneous

The numpy array is homogeneous in nature, which means that elements of the numpy array are of the same data type.

In our example below, we have silently written the last element as 1.0 and because of this, the data type of numpy array becomes float, even though all other elements were integer.

numpy_array = np.array([[1,2,3,4], [5,6,7,1.0]])
print(numpy_array)
numpy_array.dtype

[[1. 2. 3. 4.]
 [5. 6. 7. 1.]]

dtype('float64')

Python Numpy