The course aims at providing the foundations of statistical learning, machine learning, and deep learning. The presentation of specific statistical techniques and the main statistical and machine learning models will enable students to achieve the essential skills needed to conduct in-depth analyses of real-world data, with particular attention to financial data. In line with the reliability requirements for black-box models set forth by the European Commission and outlined in the regulation for trustworthy Artificial Intelligence, the course will also introduce recent statistical approaches from the literature addressed to assess the “safety” level of AI methodologies. By the end of the course, students are expected to have developed a solid mastery of both theoretical and coding concepts, allowing them to independently carry out data analysis and communicate the results using appropriate statistical language.
Course Prerequisites
In order to pass the exam, the student must have acquired an adequate knowledge of the contents covered in the mathematical-statistical area and in the coding.
Teaching Methods
The teaching methods will include: - lectures, in which the course topics will be illustrated from both a theoretical and practical perspective, with the presentation of examples. In particular, the lectures will be supported by the use of slides and the whiteboard for further explanations related to calculations and the formalization of theoretical aspects; - practical sessions, in which real data analysis will be proposed through the use of the models and statistical techniques presented during the course. The implementation will be carried out using the statistical software R.
Assessment Methods
Final oral discussion of a case study [dataset analysis and presentation of results using slides].
Texts
- Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani: An Introduction to Statistical Learning with Applications in R, II Edition, Springer (2023) - Bret Lanzt: Machine Learning with R, PUCKY Publishing open source (2013) - Scientific papers will be also recommended during the course.
Contents
The specific topics covered in the course are: 1) statistical learning models: linear regression models and logistic regression models; 2) machine learning models: decision tree models, random forest models; 3) deep learning models: neural networks models; 4) SAFE approaches for evaluating model trustworthiness; 5) regularization techniques; 6) factor models; 7) data science applications in finance.
