ID:
ECO0278
Durata (ore):
40
CFU:
6
SSD:
ECONOMIA APPLICATA
Economia applicata
Anno:
2026
Dati Generali
Periodo di attività
Secondo Semestre (15/02/2027 - 28/05/2027)
Syllabus
Obiettivi Formativi
Machine learning (ML) is a branch of Artificial Intelligence (AI) originally developed to enable computers to learn from examples and predict future events. Today, ML techniques encompass a wide range of advanced statistical methods for regression and classification, applied across many fields including medical diagnostics, credit risk modelling, image and speech recognition, and financial market analysis. The statistical methods developed in the ML literature have been particularly successful in "Big Data" settings, where data are large either in the number of observations, the number of variables, or both. This course presents Machine Learning techniques from an econometric perspective, with emphasis on both theory and practical implementation.
By the end of the course, students will be able to:
- Identify and distinguish the main statistical learning problems (regression, classification, unsupervised learning);
- Describe the theoretical foundations of the main Machine Learning methods covered in the course (linear and logistic regression, shrinkage methods, tree-based methods, neural networks);
- Apply Machine Learning methods to real-world data using the statistical software R;
- Evaluate the predictive performance of competing models using cross-validation and appropriate accuracy measures;
- Critically analyze and interpret results in an applied economic context;
- Design and conduct an independent data analysis, from exploratory analysis through to the presentation of findings.
By the end of the course, students will be able to:
- Identify and distinguish the main statistical learning problems (regression, classification, unsupervised learning);
- Describe the theoretical foundations of the main Machine Learning methods covered in the course (linear and logistic regression, shrinkage methods, tree-based methods, neural networks);
- Apply Machine Learning methods to real-world data using the statistical software R;
- Evaluate the predictive performance of competing models using cross-validation and appropriate accuracy measures;
- Critically analyze and interpret results in an applied economic context;
- Design and conduct an independent data analysis, from exploratory analysis through to the presentation of findings.
Prerequisiti
Intermediate knowledge of econometrics, statistics and linear algebra.
Metodi didattici
In-class lectures will be complemented by practical examples and group hands-on sessions using real-world data. Students are expected to actively participate in all lectures and hands-on activities.
Verifica Apprendimento
The final assessment consists of three components:
- Group project (50%): developed in pairs using the R software, in which students apply the techniques covered during the course to a real-world dataset assigned by the instructor. The project is evaluated on the basis of: correctness of the methods applied; quality of interpretation and discussion of results; clarity and readability of the written code; and consistency with the approaches, methods and programming style developed during the course lectures and hands-on sessions.
- Project presentation (30%): each group presents their work to the class in a final session of approximately 15 minutes. The presentation is evaluated on the basis of: clarity of exposition, ability to explain and justify methodological choices, and quality of responses to questions from the instructor.
- Active participation in class and discussion (20%): assessed on the basis of attendance and quality of contributions during lectures, hands-on sessions, and the final presentation day.
No midterm or partial examinations will be held for this course.
- Group project (50%): developed in pairs using the R software, in which students apply the techniques covered during the course to a real-world dataset assigned by the instructor. The project is evaluated on the basis of: correctness of the methods applied; quality of interpretation and discussion of results; clarity and readability of the written code; and consistency with the approaches, methods and programming style developed during the course lectures and hands-on sessions.
- Project presentation (30%): each group presents their work to the class in a final session of approximately 15 minutes. The presentation is evaluated on the basis of: clarity of exposition, ability to explain and justify methodological choices, and quality of responses to questions from the instructor.
- Active participation in class and discussion (20%): assessed on the basis of attendance and quality of contributions during lectures, hands-on sessions, and the final presentation day.
No midterm or partial examinations will be held for this course.
Contenuti
The course is organized in 10 lectures and will cover the following topics:
1. Introduction to Statistical Learning (4 hours): basic concepts and definitions; supervised vs unsupervised learning; regression vs classification problems; assessing model accuracy; bias-variance trade-off.
2. Linear Regression Models (4 hours): refresh, extensions and potential problems.
3. Classification (4 hours): linear methods, logistic regression, discriminant analysis.
4. Resampling Methods (4 hours): cross-validation and bootstrap.
5. Model Selection and Shrinkage Methods (4 hours): ridge regression and LASSO.
6. Moving Beyond Linearity (4 hours): polynomial regression, splines, GAM.
7. Tree-Based Methods (4 hours).
8. Unsupervised Learning: clustering and principal component analysis (4 hours).
9. Introduction to Neural Networks and Deep Learning (4 hours).
10. Hackathon and project presentations (4 hours).
1. Introduction to Statistical Learning (4 hours): basic concepts and definitions; supervised vs unsupervised learning; regression vs classification problems; assessing model accuracy; bias-variance trade-off.
2. Linear Regression Models (4 hours): refresh, extensions and potential problems.
3. Classification (4 hours): linear methods, logistic regression, discriminant analysis.
4. Resampling Methods (4 hours): cross-validation and bootstrap.
5. Model Selection and Shrinkage Methods (4 hours): ridge regression and LASSO.
6. Moving Beyond Linearity (4 hours): polynomial regression, splines, GAM.
7. Tree-Based Methods (4 hours).
8. Unsupervised Learning: clustering and principal component analysis (4 hours).
9. Introduction to Neural Networks and Deep Learning (4 hours).
10. Hackathon and project presentations (4 hours).
Lingua Insegnamento
INGLESE
Altre informazioni
The syllabus may be subject to modifications during the course. Please check the course page on the e-learning platform periodically for updates and communications from the instructor.
Office hours: students are encouraged to contact the instructor by email to schedule an appointment, or to approach the instructor directly after class.
Tutoring: https://www.uninsubria.it/servizi/tutti-i-servizi/tutorato-dieco.
This course contributes to the achievement of the following United Nations 2030 Agenda Sustainable Development Goals: Goal 4 — Quality Education; Goal 9 — Industry, Innovation and Infrastructure.
Office hours: students are encouraged to contact the instructor by email to schedule an appointment, or to approach the instructor directly after class.
Tutoring: https://www.uninsubria.it/servizi/tutti-i-servizi/tutorato-dieco.
This course contributes to the achievement of the following United Nations 2030 Agenda Sustainable Development Goals: Goal 4 — Quality Education; Goal 9 — Industry, Innovation and Infrastructure.
Corsi
Corsi
No Results Found
Persone
Persone
Ricercatori a tempo determinato
No Results Found