Machine Learning for Econometrics and Data Science

COURSE OBJECTIVE
The aim is to explore, study, and develop quantitative learning systems, methods, and algorithms with the purpose to improve the performance with learning from large data sets using powerful computers.

COURSE CONTENT
Machine learning originates from computer science and statistics with the goal of exploring, studying, and developing learning systems,
methods, and algorithms that can improve their performance with learning from data. This course is designed to provide students an introduction to the main foundations of machine learning. We adopt principles from probability (Bayes rule, conditioning, expectations, independence), linear algebra (vector and matrix operations, eigenvectors, SVD), and calculus (gradients, Jacobians) to propose a formal analysis of the performance of machine learning algorithms. Focusing on the supervised learning framework, we formalise the problem
of learning to predict based on examples. We introduce the notions of predictor, generalisation risk, Bayes risk and target function,
empirical error, model and empirical risk minimisation, learning rules, approximation and estimation errors decomposition, and derive learning guarantees under different classification and regression frameworks. We relate these notions to machine learning principles such as model selection, over-fitting, and under-fitting, and techniques such as cross-validation and regularization. In case work we implement learning algorithms and interpret the results.

TEACHING METHODS
Lectures (4 hours, each week) and Tutorials (2 hours, each week)

TYPE OF ASSESSMENT
Written exam plus an assignment