Machine Learning for Regression

Information

This curriculum module contains interactive MATLAB® live scripts that teach the basics of machine learning for regression.

Background

You can use these live scripts as demonstrations in lectures, class activities, or interactive assignments outside class. This module covers the difference between regression, classification, and clustering, as well as feature engineering and feature extraction, overfitting and underfitting, and a variety of machine learning models commonly used for regression. It also includes a detailed example of applying regression models for electricity load forecasting using real-world data.

The instructions inside the live scripts will guide you through the exercises and activities. Get started with each live script by running it one section at a time. To stop running the script or a section midway (for example, when an animation is in progress), use the Stop button in the RUN section of the Live Editor tab in the MATLAB Toolstrip.

Prerequisites

This module does not assume any prior exposure to the subject of machine learning.

Getting Started

Accessing the Module

On MATLAB Online:

Use the MATLAB Online link above to download the module. You will be prompted to log in or create a MathWorks account. The project will be loaded, and you will see an app with several navigation options to get you started.

On Desktop:

Download or clone this repository. Open MATLAB, navigate to the folder containing these scripts and double-click on MLforRegression.prj. It will add the appropriate files to your MATLAB path and open an app that asks you where you would like to start.

Ensure you have all the required products (listed below) installed. If you need to include a product, add it using the Add-On Explorer. To install an add-on, go to the Home tab and select Add-Ons > Get Add-Ons.

Products

MATLAB® is used throughout. Tools from Statistics and Machine Learning Toolbox™, Deep Learning Toolbox™, and Econometrics Toolbox™ are used frequently as well. Parallel Computing Toolbox™ is utilized specifically for the parfor function. Curve Fitting Toolbox™ is used specifically for the fittype function.

Scripts

MachineLearningIntro.mlx

	In this script, students will... $\bullet$ Learn the difference between regression, classification, and clustering $\bullet$ Define feature engineering/extraction $\bullet$ Identify and use different machine learning models commonly used for regression $\bullet$ Be able to explain overfitting and underfitting

LoadForecastRegression.mlx

	In this script, students will... $\bullet$ Apply the machine learning workflow to solve a problem in time series forecasting $\bullet$ Engineer appropriate features to solve the forecasting problem $\bullet$ Validate and compare different types of regression models $\bullet$ Test and evaluate the trained model to make predictions

FE1_ProgrammaticML.mlx and FE2_LoadForecastDL.mlx

	In these scripts, students will... $\bullet$ Expand on the practical problem presented in LoadForecastRegression.mlx $\bullet$ Define feature engineering/extraction $\bullet$ Identify and use different machine learning models commonly used for regression $\bullet$ Be able to explain overfitting and underfitting