Course Syllabus
Aims
Here are the main goals of this class:
- Knowledge and understanding
To provide students with a solid grounding in statistical methods and data analysis techniques commonly used in physics, including the conceptual foundations necessary to critically interpret experimental data.
- Applying knowledge and understanding
To develop the ability to apply statistical tools and data analysis methods to real-world physics problems, through hands-on exercises and interactive sessions aimed at reinforcing practical understanding.
- Making judgements
To cultivate critical thinking in the evaluation of data quality, in the choice of appropriate statistical models, and in the interpretation of outcomes within the broader context of physical theories and experimental limitations.
- Communication skills
To strengthen students’ capacity to clearly and effectively present their data analysis process and results, using written reports, oral discussions, and visual tools tailored to both scientific and interdisciplinary audiences.
- Learning skills
To encourage autonomous learning and the continuous development of data analysis skills, enabling students to confidently approach increasingly complex datasets and keep pace with evolving techniques in physics research.
Contents
- Recap of fundamental concepts of probability and statistics
- Recap of parameters estimation
- Hypothesis Testing
- Confidence interval estimation
- Intro to MC methods
- Intro to Machine learning and Deep Learning
- Unfolding
- Analysis workflow Walk-Through
Detailed program
Intro and recaps
- Intro to fundamental concepts of probability and statistics
- Recap of parameters estimation (Maximum Likelihood, Least Squares, Bayesian approach)
Hypothesis Testing
- From basic concepts to P-values determination (P-values, GoF, Nuisance parameters, Look elsewhere effect)
Interval Estimation
- Frequentist Methods
- Bayesian Methods
- Coverage and Sensitivity
Intro to MonteCarlo methods
- Random number generation
* Accept-Reject method and other algorithms
* Markov Chain MonteCarlo
Intro to machine learning
- Fundamental concepts (loss, generalization)
- Supervised learning (regression, classification)
- Unsupervised learning (auto-encoders, clustering)
- Dimensional reduction
- MLP, BDT, SVM, PCA, …
Deep Learning
- Intro to Convolutional Networks, Generative Adversarial Networks, Graph Neural Networks
Unfolding
- Inverse Problem
- Regularisation Methods
- ML based unfolding
Analysis example end-to-end
- examples of analysis flows and presentation of results (examples from multiple fields)
- systematic uncertainty (estimation and treatment)
- Incorporating uncertainty in ML
Prerequisites
A basic knowledge of programming (preferably Python) is preferred. Material will be provided before the beginning of the course for those who don’t have programming experience.
Teaching form
Frontal lectures and hands on exercises. Students are required to bring their own laptop.
Semester
Second semester
Assessment method
Each student is required to develop a data analysis project (subject to be agreed with the teacher) and that will serve as a starting point for the oral exam. The oral exam will also include questions on all the topics seen during the lectures.
Office hours
Contact us via email to book an appointment.
