Course contentsShow
← Data Science
Lesson 595 of 2,14516. Multiple Linear RegressionFree lesson

From Simple to Multiple Linear Regression

Extend the simple linear regression framework to include multiple predictor variables simultaneously.

From Simple to Multiple Linear Regression

What you'll learn: How to expand from one predictor variable to many, allowing you to model more complex real-world relationships.

The Core Idea

Until now, you've worked with simple linear regression: predicting an outcome using a single predictor variable (like predicting salary from years of experience). But real-world phenomena rarely depend on just one factor. Multiple linear regression extends the same least squares framework to include several predictor variables simultaneously.

Think of it like upgrading from a one-dimensional thermometer to a multi-sensor weather station. Instead of just temperature, you now measure humidity, wind speed, and pressure—all contributing to your weather forecast.

The Mathematical Extension

Simple regression uses:

y = β₀ + β₁x + ε

Multiple regression extends this to:

y = β₀ + β₁x₁ + β₂x₂ + ... + βₚxₚ + ε

Each predictor x₁, x₂, ..., xₚ gets its own coefficient β₁, β₂, ..., βₚ, representing its unique contribution while holding all other predictors constant. The intercept β₀ still represents the expected value when all predictors equal zero.

Why This Matters

Multiple regression allows you to:

  • Control for confounding variables you learned about in partial correlation
  • Model realistic scenarios (house prices depend on square footage, bedrooms, location, age...)
  • Isolate the effect of each predictor independently
  • Improve prediction accuracy by capturing more variation in the outcome

The same assumptions, diagnostics, and least squares principles still apply—they just operate in higher-dimensional space. You'll still minimize residuals, check for linearity and homoscedasticity, and interpret R-squared, but now with multiple predictors working together.

Key Takeaway: Multiple linear regression extends simple regression from one predictor to many, using the same least squares framework to estimate how each variable independently contributes to the outcome.