Skip to content

03 Regression

Preparation

Ch 4 (except "Logistic Regression")

Material

Session material: In this folder.

Hitters.csv

Regression - Empty (PDF)

Regression X (PDF)

Regression Y (PDF)

Regression - Hitters (Jupyter Notebook)

Lecture

Session Description

Today, we are going to look at regression algorithms, where instead of predicting a class you predict some continuous variable.

We also discuss what is meant by “regularization” and consider the R² performance metric for regression.

Key Concepts

  • Ordinary Least Squares (OLS) regression
  • Ridge regression
  • Lasso regression

Learning Objectives

  • Explain what is meant by "regression" and in which contexts it applies
  • Explain the following linear regression models, their strengths and weaknesses, and apply them in python:
    • Ordinary Least Squares (OLS) regression
    • Ridge regression
    • Lasso regression
    • Elastic Net Regression
  • Explain what is meant by the term "regularization" in an ML-context
  • Describe what is meant by "bias" and "variance" in relation to ML-algorithms
  • Explain the R²-metric for evaluating the performance of a linear regression algorithm
  • Explain the MSE metric

Video Lectures

3.1 Linear Regression

3.2 Regularization

In general we recommend Alexander Ihler’s course “Machine Learning and Data Mining”.