Why & Where Matrix used in Real Life Data Science and Machine Learning?

Uses of Matrices in Data Science and Machine Learning

In data science and machine learning, we deal with a lot of numbers and datasets ( dataset means Collection of Multiple tables).

Matrices help us to handle large numbers easily and quickly.


Here are the main uses of matrices:


Storing Data

Any dataset you see (Excel table, CSV file, SQL table) is like a matrix.

Rows = records or samples. ( Row 1, 2, 3, etc.. )
Columns = features or variables. ( Column A, B, C etc...)

Column contains Name, Subject, Employee ID, Attendance, Department, Height, Weight - Such Headings. So Columns contain Features.

Row contains the List of Names, like Akash, Ajith etc.. List of Subject like Maths, English etc.. and Height values like 5 feet, 6 feet etc..  So Row Contains Values.

This method of Storing Data is Best for Everything! (Both Humans and Computers)





Image Processing


Digital Images are stored as matrices. Each Matrix Numbers represent Color or density of a Single Pixel of image.

For example, a black and white image is a 2D matrix of pixel values. A color image is 3 matrices (Red, Green, Blue).



Here you can see Image of Panda is made up of thousands of Small Pixels. These Pixels all work together and show us a Single Image.

Each Pixel has a Matrix Value that tell Computer what color to show in that Pixel.


Linear Regression and Algorithms


Machine learning algorithms like Linear Regression, Logistic Regression, SVM, and Neural Networks use matrices for training and predictions.

Linear regression and many machine learning algorithms use matrices because they allow fast and efficient calculations when working with large data.

Converting Inputs and Outputs of Linear Regression into Matrix Form helps Computer to do mathematical calculations quickly using Python NumPy Libraries.

Weights × Inputs = Prediction. This is a core step in most models. Matrix multiplication helps to do it faster using NumPy or other libraries.





Neural Networks


Each layer in a neural network uses matrix operations (dot product, addition, activation). Without matrices, training a neural network is not possible.

In neuron network, each Neuron or Layer receives input data like Name, Subject, Attendance, Behavior etc. in a matric Number format ( 2121, 23, 100, 45) like this.

Then it performs Matrix multiplication to weight the data ( to find the Importance of data)

Then it send to activation room. The activation room decides whether to activate that neuron or not, so it can provide Correct answer as Output.




Consider Each Silver box outside as Neuron or Layer. These layers each contain a Row value. (1243,23,4545,676) something like this.

Now, the data is sent from Layers to Inner silver Cylinders. They do matrix Multiplication and find the Importance or weight of each data.

Last, it is sent to small colorless Main Round Cylinder, Where it decides whether this input data should be considered for giving proper answer or not. 


Transformations


In data science, we transform data (scale, rotate, normalize). These are done using matrix operations.

We transform the data before giving it to Models. We use Scaling, Rotation and Normalization to transform the data for better accuracy and Speed.

We need Scaling Formula, Rotation Matrix for images etc. 

Using Matrix Transformation, Image can be scaled Rotated, Flipped without affecting Image structure.





Speed and Efficiency


Matrix math is faster and easier for computers than normal number-by-number calculations. That’s why big models depend on matrix algebra.


Example:

Normal Math (one-by-one calculation):

Let’s say you want to calculate salary based on hours worked and rate per hour.

For 3 employees:

Employee 1: 8 hours × 100 = 800
Employee 2: 9 hours × 100 = 900
Employee 3: 7 hours × 100 = 700

You are doing multiplication one by one.


Matrix Math (all-at-once calculation):

Now instead of doing 3 calculations separately, we can do it using matrices.

Hours = [8, 9, 7]
Rate = 100

Now you multiply: [ 8 9 7 ] * [100] = [ 800, 900, 700 ]

Computer does all three calculations in one step using matrix multiplication. This is much faster, especially when you have thousands or millions of rows

Comments

Post a Comment

Popular posts from this blog

What is Artificial Intelligence? What is Machine Learning? What is Data Science? how they are related to each other?

Image
Artificial Intelligence means making Computers think like Human Brain, choosing correct & realistic decisions, answers based on the Current circumstance and importance. Personal Example: Data Science = What is Actually Happening in your Life. Machine Learning = What you are learning and experiencing in your Life. Artificial Intelligence = Taking Correct Decisions based on your Life Experience Another Example: You are AI Engineer & you created a Personal AI Friend using Python. Now, You give all information about your Financial Status, Economic Status, Emotional Status, Social Status, Current Situation in your Life, Current Situation in your Family, Current Situation in your Country etc etc... and your Current problems in an Excel file or SQL Database. Now you are giving this data to your AI tool which you created. 1. It will analyze the information, choose which is important and which is useless based on your current situation. 2. Now, it sends that important Information to Mac...
Read more

Linear Algebra - What is it?

Linear Algebra is a branch of mathematics that deals with vectors , matrices , and linear transformations . It's one of the foundations of data science, machine learning, and computer graphics. In Simple Words : Linear algebra helps us: Solve equations with multiple variables ( variables - a value, number or anything that changes anytime eg, Mark, Temperature, Price ) Represent and transform data using matrices ( Matrix - numbers arranged in rows and columns) Understand how systems behave (like how multiple inputs affect an output) Key Concepts : Scalars – Scalar means Magnitude(Speed). So it is just a single number (e.g., 8 or 6 or 10 etc..) Vectors – Vector means Both Magnitude (Speed) and Direction. So it is a ordered list of numbers (e.g., [2, 3] or [4, 5] or [10, 50] etc..) Matrices – A grid of numbers arranged in rows and Columns  (e.g., a 2x2 matrix or 4x4 matrix) Matrix Multiplication – Combining 2 or more matrices to transform data. Linear Transformation - Multiply...
Read more

What is a Python Library?

When we use Python for data analysis, machine learning, or any programming, we do not write each and every code. Why Should we? Some experts already wrote code for us and packed it as a Library. This reusable collection of code is called a Python Library . A library in Python is like a ready-made toolbox. Instead of building every tool yourself, you take the tools from the toolbox and start using it in your code. Why do we need Python Libraries? Save time – We do not write the same code again and again. Easy to use – Just install, import, and start using. Powerful features – Some libraries can do complex tasks in one line of code. Community support – Big communities maintain them, so they keep improving. Examples of Python Data Science Libraries: Pandas – Works with data tables like Excel, filtering and summarizing data NumPy – Works with numbers, arrays, mathematical calculations Matplotlib – Creates graphs and charts Scikit-learn – For machine learning models like regres...
Read more