Linear Regression in MATLAB: A Complete Guide

Introduction to Linear Regression in MATLAB

Linear regression is a fundamental statistical technique used to model the relationship between a dependent variable and one or more independent variables. MATLAB, a powerful numerical computing environment, offers multiple functions to perform linear regression efficiently. Whether you're a beginner or an experienced data analyst, understanding how to implement linear regression in MATLAB is essential for data analysis, forecasting, and predictive modeling.

In this comprehensive guide, we'll explore the most common MATLAB functions for linear regression: fitlm, regress, and polyfit. We'll provide step-by-step examples and best practices to help you master linear regression in MATLAB.

Why Use MATLAB for Linear Regression?

MATLAB provides a rich set of tools for statistical analysis, making it a preferred choice for engineers, scientists, and researchers. Key advantages include:

• Ease of use: Functions like fitlm and regress simplify complex computations.
• Visualization: Built-in plotting functions help visualize regression results.
• Integration: Seamlessly integrate with other MATLAB toolboxes for advanced analysis.
• Performance: Optimized for large datasets and matrix operations.

Methods for Linear Regression in MATLAB

MATLAB offers several functions for linear regression, each suited to different scenarios. Let's explore the three most popular ones: fitlm, regress, and polyfit.

1. Using fitlm for Linear Regression

The fitlm function is part of the Statistics and Machine Learning Toolbox. It fits a linear model to data and returns a LinearModel object with extensive diagnostic information. It's ideal for simple and multiple linear regression.

Syntax:

mdl = fitlm(X, y)

Example: Suppose we have a dataset with one predictor (X) and one response (y).

X = [1; 2; 3; 4; 5];
y = [2.1; 3.9; 6.2; 8.1; 9.8];
mdl = fitlm(X, y);
disp(mdl)

The output includes coefficients, R-squared, p-values, and more. You can also plot the fit:

plot(mdl)

Advantages:

• Comprehensive statistical summary.
• Supports multiple predictors and categorical variables.
• Easy to interpret and visualize.

2. Using regress for Multiple Linear Regression

The regress function performs multiple linear regression using least squares. It returns coefficient estimates, confidence intervals, residuals, and more. It's part of the base MATLAB and doesn't require additional toolboxes.

Syntax:

[b, bint, r, rint, stats] = regress(y, X)

Example: For multiple regression with two predictors:

X = [ones(5,1), (1:5)', (2:6)']; % Include intercept
y = [2.1; 3.9; 6.2; 8.1; 9.8];
[b, bint, r, rint, stats] = regress(y, X);
disp(b); disp(stats);

Advantages:

• No toolbox required.
• Provides detailed statistical outputs.
• Fast for large datasets.

Note: You must manually add a column of ones for the intercept.

3. Using polyfit for Polynomial Regression

The polyfit function fits a polynomial curve to data. For linear regression, set the degree to 1. It's simple and effective for basic trend analysis.

Syntax:

p = polyfit(x, y, 1)

Example:

x = [1; 2; 3; 4; 5];
y = [2.1; 3.9; 6.2; 8.1; 9.8];
p = polyfit(x, y, 1);
y_fit = polyval(p, x);
plot(x, y, 'o', x, y_fit, '-')

Advantages:

• Simple and quick for basic linear fits.
• Works well for polynomial regression of any degree.
• No additional toolboxes needed.

Choosing the Right Function

Selecting the appropriate function depends on your needs:

• fitlm: Best for comprehensive statistical analysis and when you need detailed diagnostics.
• regress: Ideal for multiple regression without toolboxes, especially when you need confidence intervals and residuals.
• polyfit: Perfect for simple linear or polynomial regression with minimal output.

Practical Tips for Linear Regression in MATLAB

• Check assumptions: Ensure linearity, independence, homoscedasticity, and normality of residuals.
• Standardize data: For multiple regression, consider standardizing predictors to compare coefficients.
• Validate model: Use cross-validation or hold-out sets to assess predictive performance.
• Visualize: Always plot data and fitted line to detect outliers or nonlinear patterns.

Conclusion

MATLAB provides robust tools for linear regression, catering to various levels of complexity. Whether you use fitlm for detailed analysis, regress for multiple regression, or polyfit for quick fits, you can efficiently model relationships in your data. By mastering these functions, you'll enhance your data analysis capabilities and make more informed decisions. Start practicing with the examples above and explore MATLAB's documentation for advanced features like robust regression and regularization.