Key Module Information

Module Code: ECOM193

Module Title: Statistical Machine Learning in Finance

Credit Value: 15

Level: 7 (MSc)

Semester: B (2)

Prerequisites:  Ideally, students should have taken an elementary course in statistics            previously. Some knowledge of linear algebra and a little high school calculus would also be useful. Prior knowledge of the R free software environment would be advantageous but is    not essential as lecture and support class guidance will be given.

Module Organiser

Name: Dr Richard Saldanha

Office Location: External Visiting Lecturer (contact via email)

Office hours: 15:00 Tuesdays (immediately after lecture)

Email: [email protected]

Module Delivery

Each week you will be expected to engage with material and exercises (independent learning activities) posted on the module’s QMplus page, alongside attending and     participating in scheduled teaching activities.

Details of the scheduled teaching activities will be in your timetable. This can be accessed via the My Timetable option in QMplus.

Lecture

Lecture on Tuesdays 13:00– 15:00 • Geography 2.26

Classes

Details of your class allocation can be found in MyTimetable on QMplus.

Module Content

Module Aim

The aim of this module is to give students a good understanding of the foundations of      statistical machine learning and the use of selected methods in finance. There is a strong emphasis on practical application using R to model the provided datasets.

Learning Outcomes

Upon completion of this module, students should have a good grasp of statistical machine    learning and be able to undertake their own ML projects in a relatively unsupervised manner.

Module Outline

Suggested background/supplementary reading in square brackets where applicable; please refer to the Reading List given in the next section.

Introduction

Course outline; what is machine learning? supervised versus unsupervised learning; the difference between classification and regression; installing and using R. [ISLR Ch.1]

Lecture 1: Machine Learning Essentials

Essential statistical knowledge; model accuracy; bias-variance tradeoff; resampling methods. [ISLR Ch.2&5]

Lecture 2: Linear Regression

Regression basics: model fitting and diagnostics. [ISLR Ch.3]

Lecture 3: Classification

Logistic regression; discriminant analysis; naïve Bayes; nearest-neighbour classification. [ISLR Ch.4]

Online Multiple Choice Test 1 (10%)

Lecture 4: Linear Model Selection and Regularization

Automatic model selection methods; regularization and shrinkage; dimensionality reduction via principal component analysis. [ISLR Ch.6]

Lecture 5: Towards Non-Linear Models

Polynomial regression; step functions; basis functions; regression splines; local regression; generalized additive models. [ISLR Ch.7]

Midterm Coursework Assessment (20%)

Lecture 6: Tree-Based Methods

Decision trees; bootstrap aggregation (bagging); random forests. [ISLR Ch.8]

Lecture 7: Support Vector Machines

Using support vector machines to solve classification problems. [ISLR Ch.9]

Online Multiple Choice Test 2 (10%)

Lecture 8: Neural Networks and Deep Learning

An explanation of artificial neural networks and their uses. [ISLR Ch.10]

Lecture 9: Reinforcement Learning

A brief look at reinforcement learning as an ML technique.

Lecture 10: Optimization

The role of optimization in machine learning; identifying the optimization problem; typical optimization approaches; gradient boosting; guidance on tackling the assessed project.

Online Multiple Choice Test 3 (10%)

Project Assessment (50%)

Reading List

[ISLR] James, G., Witten, D, Hastie, T. and Tibshirani, R. (2021) An Introduction to Statistical Learning: with Applications in R . Second edition. Springer-Verlag, New York.

ISLR is the main textbook for the course. It also focuses on undertaking analyses  using R. The book can be used as a complementary guide to the ML methods introduced in lectures. Module topics are covered largely in ISLR chapter order (Lectures     1–8). Course examples and datasets are, however, quite different to ISLR as they   are tailored to finance. The book is freely available online.

[ESL] Hastie, T., Tibshirani, R. and Friedman, J. (2009) The Elements of Statistical Learning: Data Mining, Inference and Prediction. Second edition. Springer-Verlag, New York.

ESL is a popular book detailing nearly all of the topics contained in the module. It generally covers topics in a little more depth than ISLR. A corrected 12th    printing of the entire book as of January 2017 is available online.

[MSDA] Rice, J.A. (2006) Mathematical Statistics and Data Analysis. Third edition. Brooks Cole, Belmont, CA.

Students may find this book useful for statistical concepts such as bias, variance

and likelihood.

[CASI] Efron, B. and Hastie, T. (2016) ComputerAge Statistical Inference: Algorithms, Evidence, and Data Science. Cambridge University Press, New York.

For students wishing to delve into the subject area more deeply. CASI describes the revolution in data analysis following the introduction of electronic computation in the 1950s. Many of the module topics are covered by CASI but it is a harder more   technical read than both ISLR and ESL.