📘 Section 12.1.1 – Multiple Linear Regression Model

Why Do We Need Multiple Linear Regression?

In the real world, outcomes rarely depend on a single variable. Multiple linear regression (MLR) lets us examine how several factors simultaneously affect a response variable. This is important for making accurate predictions and understanding complex systems.

What Is It Used For?

MLR is used in engineering, business, healthcare, and social sciences to:

Why Not Use Single-Factor Models?

Single-variable models oversimplify reality. For example, predicting a student's GPA using only study hours ignores other important factors like sleep, stress, and class attendance. MLR accounts for multiple influences simultaneously.

Single vs. Multiple Regression – Key Differences

🎓 Student-Related Example

Suppose we want to predict a student’s GPA (Y). Possible predictors:

GPA = β₀ + β₁x₁ + β₂x₂ + β₃x₃ + ε

This MLR model helps quantify how each behavior affects GPA while holding others constant.

Introduction

Multiple linear regression involves more than one predictor variable. The general form of the model is:

Y = β₀ + β₁x₁ + β₂x₂ + ... + βₖxₖ + ε

This equation describes a hyperplane in (k+1)-dimensional space. Each coefficient βⱼ indicates the effect of one predictor holding the others constant.

Basic Example

Vehicle mileage depends on weight (x₁) and engine displacement (x₂):

Y = β₀ + β₁x₁ + β₂x₂ + ε

Example model: E(Y) = 50 + 10x₁ + 7x₂

Polynomial Models

Higher-order polynomial regression can also be expressed in linear regression format:

Y = β₀ + β₁x + β₂x² + β₃x³ + ε

Rewriting using new variables x₁ = x, x₂ = x², x₃ = x³:

Y = β₀ + β₁x₁ + β₂x₂ + β₃x₃ + ε

Interaction Terms

Interaction effects are represented by cross-product terms. For example:

Y = β₀ + β₁x₁ + β₂x₂ + β₁₂x₁x₂ + ε

This model accounts for effects that vary depending on the level of another predictor.

Second-Order Model with Interaction

Flexible model that includes quadratic and interaction terms:

Y = β₀ + β₁x₁ + β₂x₂ + β₁₁x₁² + β₂₂x₂² + β₁₂x₁x₂ + ε

Can be rewritten with new variables:

Y = β₀ + β₁x₁ + β₂x₂ + β₃x₃ + β₄x₄ + β₅x₅ + ε

Where: x₃ = x₁², x₄ = x₂², x₅ = x₁x₂

📊 Visualization Placeholder

This section can include regression plane plots and contour plots using 3D graphics or Chart.js projections (coming soon).