Home / Blog / Data Science Digital Book / Mathematical Foundations

# Mathematical Foundations

• July 15, 2023
• 3489
• 20 ### Meet the Author : Mr. Bharani Kumar

Bharani Kumar Depuru is a well known IT personality from Hyderabad. He is the Founder and Director of Innodatatics Pvt Ltd and 360DigiTMG. Bharani Kumar is an IIT and ISB alumni with more than 17 years of experience, he held prominent positions in the IT elites like HSBC, ITC Infotech, Infosys, and Deloitte. He is a prevalent IT consultant specializing in Industrial Revolution 4.0 implementation, Data Analytics practice setup, Artificial Intelligence, Big Data Analytics, Industrial IoT, Business Intelligence and Business Management. Bharani Kumar is also the chief trainer at 360DigiTMG with more than Ten years of experience and has been making the IT transition journey easy for his students. 360DigiTMG is at the forefront of delivering quality education, thereby bridging the gap between academia and industry.

### Basic Matrix Operations:

• Matrix Transpose
• Matrix Multiplication
• Matrix Inverse
• Orthogonal Matrix
• Eigenvectors and Eigenvalues

## Dimension Analysis and Factor Analysis (PCA, SVD, LDA)

Features, variables, and columns are other terms for dimensions.

Dimensionality reduction is the process of extracting features from input variables that are composed of hundreds of variables.

Less dimensions provide rapid computations and clear interpretation, which reduces overfit situations and helps to avoid collinearity.

The ability to visualise multivariate data in a 2D space is another advantage of dimensionality reduction.

In this blog, among the various methods accessible

we will discuss the most popular methods:

## PCA - Principal Components Analysis

PCA is used to analyse dense data, which is data that is quantitative in nature and does not include many zeros.

Principal Component Analysis (PCA) is used to divide a large number of characteristics into an equal number of features known as Principal Components (PCs).

The initial group of PCs alone can gather the most information, but these PCs capture 100% of the data.

With minimal information loss, PCA enables us to considerably reduce the dataset size. Applying PCA is ineffective if the initial dataset contains characteristics that are all associated.

Each PC will record all the data that is present in the original dataset's variables.

## Benefits of PCA

• Reduction of number of features & hence faster processing.
• Identify the relationship between multiple columns at one go by interpreting the weights of PCs.
• Visualizing multidimensional data using a 2D visualization technique.
• Inputs being correlated is called as collinearity and this is a problem, which is overcome by PCA because it makes the inputs uncorrelated.
• Helps in identifying similar columns.

The ith principal component is a weighted average of original measurements / columns: Weights (aij) are chosen such that:

• PCs are ordered by their variance (PC1 > PC2 > PC3, and so on)
• Pairs of PCs have correlation = 0
• For each PC, sum of squared weights = 1 (Unit Vector)

Data Normalization / Standardization should be performed before applying PCA.

## SVD - Singular Value Decomposition

Sparse data, or data with many items that are zeros, is reduced using a technique known as singular value decomposition, or SVD.

SVD is used to process photos and greatly aids in image processing by reducing the size of the images.

In the recommendation engine, SVD is frequently utilised.

It is a matrix decomposition method, represented as:

• Diagonal matrix (d) values are known as the singular values of the original matrix X.
• U matrix column values are called the left-singular vectors of X
• V matrix column values are called the right-singular vectors of X ## LDA - Linear Discriminant Analysis

For data with more features, dimensionality reduction is solved using linear discriminant analysis (LDA).

Considering the class label, linear discriminant analysis is a supervised algorithm.

Each class of datapoints has a centroid determined via LDA.

LDA determines a new dimension based on centroids in a way to satisfy two criteria:

• Maximize the distance between the centroid of each class.
• Minimize the variation (which LDA calls scatter and is represented by s2), within each category.

Click here to learn Data Science Course, Data Science Course in Hyderabad, Data Science Course in Bangalore

### Data Science Training Institutes in Other Locations   