Machine Learning: Regression

Welcome to the introductory module, where you will gain an overview of the course structure and the essential software tools required. The module also provides insights into the assumed background and the community aspect of the learning experience.

Delve into the fundamentals of simple linear regression, exploring the task, model, and optimization objectives. By learning techniques such as gradient descent and the impact of high leverage points, you will lay the foundation for building regression models.

Multiple regression introduces concepts such as polynomial regression and seasonality modeling. With a focus on matrix notation and gradient descent, learners will gain a deeper understanding of implementing multiple regression models for predicting house prices.

Assessing performance module provides insights into training and generalization error, overfitting, and the bias-variance tradeoff. By exploring polynomial regression and understanding the sources of error, learners will gain valuable insights into assessing model performance.

Explore the symptoms of overfitting and the balancing act in ridge regression. With a detailed focus on the ridge objective, coefficient path, and tuning parameters through cross-validation, learners will gain practical experience in implementing ridge regression using gradient descent.

Engage in the feature selection task and understand the complex algorithms for selecting features. You will explore the lasso objective, coordinate descent, and the practical considerations for choosing penalty strength. Implementing LASSO using coordinate descent will provide valuable hands-on experience.

Nearest Neighbors & Kernel Regression module delves into the limitations of parametric regression and explores the approach of k-nearest neighbors regression. By predicting house prices using k-nearest neighbors regression, learners will gain practical insights into the nuances of non-parametric regression techniques.

Closing Remarks module offers a comprehensive recap of the essential concepts covered throughout the course. It provides a consolidated overview of simple and multiple regression, assessing performance, ridge regression, feature selection, lasso, and nearest neighbor regression.