Causality

Course manual 2020/2021

Course content

Many questions in science are of a causal nature. But how can we formalize the notion of causality? How to reason about cause and effect formally? How can we discover causal relations from data? How to predict the consequences of  actions? How do causal predictions differ from ordinary predictions in statistics? This course will address all these questions, making use of the
mathematical framework of structural causal models.

Topics addressed will be causal modeling (definition of structural causal models, marginalization, confounders, selection bias, feedback loops, causal graphs, interventions, Markov properties, connections with equilibrium states of dynamical models, time dependence), causal reasoning (intervention variables, do-calculus, counterfactuals, covariate adjustment, back-door criterion, identifiability), and causal discovery (randomized controlled trials, local causal discovery, Y-structures, the PC algorithm, the FCI
algorithm). Practical exercises and computer lab exercises are offered to allow the student to practice with the material.

Study materials

Syllabus

• Lecture notes will be written during the course

Practical training material

• Exercises will be made available during the course

Objectives

• Modeling a system using a structural causal model
• Understanding the interpretation of the graph of a structural causal model
• Understanding the causal notions of confounders, selection bias, feedback loops, intervention variables
• Calculating the observational distribution, interventional distriutions and counterfactual distributions induced by a structural causal model
• Understanding and applying the Markov property for structural causal models
• Understanding and applying the causal do-calculus
• Understanding and applying covariate adjustment criteria (back-door criterion and generalizations)
• Understanding the (lack of) identifiability and its consequences for estimation from data
• Understanding and proving simple properties of basic causal discovery algorithms (RCTs, LCD, Y-structures)
• High-level understanding of the PC and FCI algorithms
• Implementing and evaluating simple causal discovery and prediction algorithms

Teaching methods

• Lecture
• Self-study

Learning activities

Attendance

This programme does not have requirements concerning attendance (TER-B).

Assessment

More details on the examination will be provided during the course.

Fraud and plagiarism

The 'Regulations governing fraud and plagiarism for UvA students' applies to this course. This will be monitored carefully. Upon suspicion of fraud or plagiarism the Examinations Board of the programme will be informed. For the 'Regulations governing fraud and plagiarism for UvA students' see: www.student.uva.nl

Course structure

Timetable

The schedule for this course is published on DataNose.

Contact information

Coordinator

• prof. dr. J.M. Mooij