Numpy For Beginners(Part 1)

Hello Everyone,

I’m Promise, a medical doctor transitioning into the world of data analytics. As I continue my learning journey, I’ve decided to share my notes with you all. My hope is that these insights will be helpful, especially for those who are also navigating the exciting field of data science.

I’ve been learning Python and recently reached the topic of NumPy. Since this is where my journey took an interesting turn, I thought it would be a great place to start sharing my notes.

So let's get into the meat of the matter:

What is NumPy?

NumPy is a Python library that stands for Numerical Python. It’s a powerful package designed to handle numerical operations efficiently. Created by Travis Oliphant, NumPy is a foundational tool in the Python data science ecosystem.

Why Do We Need NumPy?

In Python, we use variables to store data. However, a variable can only hold a single data point (or value). To store multiple data points, we often use Python lists. The challenge with lists is that they aren’t always the best choice when dealing with large datasets or performing complex numerical operations.

This is where NumPy comes in. Here’s why NumPy is preferred over lists:

• Speed: NumPy is significantly faster than lists when executing operations.
• Memory Efficiency: It uses less memory compared to lists.
• Multidimensional Arrays: NumPy makes it easy to work with multidimensional arrays (e.g., 2D, 3D, and beyond).
• Foundation for Other Libraries: Many popular data science libraries, such as Pandas, Scikit-Learn, and TensorFlow, are built on top of NumPy

How to Create a NumPy Array

To have access to numpy, you have to first import the numpy module, which is done with the code:

Import numpy as np

(While it's not compulsory to use the alias np, it is universally accepted so that everyone can understand what is being referred to)

To create a numpy array, you then use the array function in the numpy module

i.e ?np.array()

E.g

array1= np.array([7, 14, 21, 28, 35])

Here array1 represents the variable that stores the numpy array being created

Did you notice you need to use a square bracket? Why? That’s because you’re passing a list of values to the array() function.

The above is an example of a 1 dimensional(1D) array. It contains just one list

You can also create multidimensional array

?Let’s see how to create that:

-2D array:

This is like creating a list of list i.e, a list that contains other lists.

This means there is a master list which now contains other lists separated by a comma

Using a real world example, if you have a list of items in a grocery store grouped by category and each category is a different list

Imagine this as numbers because of course numpy is numerical

-In order to create a 2D array there has to be a square bracket that encloses the other lists and a comma separating each

To check the dimension of your array, use the property?? .ndim

This shows us that it is a 2D array.

Additionally if you want to see the shape of the array (number of rows and columns, you use the? property .shape

To see the size(i.e number of elements in the array), you use the property? .size

3D Arrays

A 3D array takes things a step further.

It is like a list in a list in another list ( a list that contains a list of list or a 2D array inside another list, if that makes sense)

Going back to our grocery store example: If you have a list that contains the list of items in different grocery stores by category

Here, the first level represents different stores, the second level represents categories (fruits/vegetables), and the third level contains actual items.

Of course, numpy deals with numbers and this is just an explanation of how the dimension works

To create a 3D array, there has to be

• An outer square bracket that encloses everything.
• Inside, multiple 2D lists, each enclosed in square brackets.
• Each 2D list is separated by a comma.

Other ways of creating an array

1.An array that contains zeros

You use np.zeros, then include the shape(i.e number of rows and columns you are interested in)

E.g

One dimensional:

2 Dimensional:

This created? a 2D array with 4 rows and 6 columns, however you can notice that it created them as float.

To confirm this , you can use the .dtype property

To create an array with int data type instead, you write dtype = int, while creating the array as seen below

Well, I tried to experiment,I noticed that this only works for zeros(np.zeros) and ones(np.ones), meaning you can’t create an array that contains 2, 3 , 4 or other digits this way just 0 and 1. However, there is another function that can be used to create this which will be discussed next

?

3Dimensional:

Here you have to specify the number of lists, the number of rows and the number of columns

To create a 3D list containing 1 with 3 set of lists, 2 rows and 3 columns per list

2. To create an array with fixed values other than 0 and 1, we can use the np.full function, indicating the shape and the number interested in

E.g np.full([2,3], 42])

This creates a 2D array with 2 rows and 3 columns with the number 42

?

3. You can also create a range of numbers using the arange function

E.g To create a numpy array with numbers ranging from 1 to 28 (note that 28 will not be a part of this result, it will stop at 27, don’t ask me why, that’s how it was programmed I guess!!)

This is a 1D array, however, you can convert it to either a 2D or any other dimension with the condition that the shape must align to fit the dimension you want to create .

?How do you confirm this?

E.g for 2D, you can do 3 multiplied by 9, gives you 27 which is 3 rows and 9 columns

However, you cant reshape it into a 2D array with 2 rows and 9 columns because this will only fit in 18 values instead of the required 27 and will therefore throw an error

To reshape it into a 3D array , you can do 3 x 3 x 3, which will give 3 sets of lists with 3 columns and 3 rows each

The simple trick is to do a multiplication:

Our array contains 27 values:

3 x 9 = 27, it can be shaped into a 2D array with 2 rows and 9 columns

2 x 9 = 18, it cant be shaped into a 2D array with 2 rows and 9 columns

3 x 3 x 3 = 27, it can be shaped into a 3D array with 3 sets of list with 3 rows and 3 columns each

?

4. ?You can also create a range of numbers with equal intervals between them using the arrange function but you’ll need to specify the interval

E.g a range of 10 to 90 with an interval of 5

5.Also, another way of creating a numpy array is by using the linspace function, this helps create a range of a specific number with equal intervals

E.g If you want to create a range between 1 to 10, and you want this range to contain 8 numbers with equal intervals.

Definitely the answer will be float values because you can't get an equal interval of 8 numbers between 1 and 10 as integers

To do this, in addition to the range of numbers, you also include how many values you are interested in, in this case, its 8 values.

If you've followed along to this point, I really hope this was worth your time and you benefitted one or two things from this read

This is just the beginning of my python notes journey, kindly follow along for more

Also, i'll appreciate your comments as well as feedbacks (It's a journey of learning and growth!)