Textbooks
The followings are useful textbooks for the course.
Tentative Schedule
DateTopicReadings
09/12/2022Introduction and Motivation
We start the course by looking at applications where statistical association cannot fully capture the underlying data-generating processes. We then introduce randomized controlled trials and observational studies to understand the difference between association and causation. We will use the potential outcome framework to quantify the causal effects in a simple binary case.
CIWI 1, 2.1-2.2, 3.1-3.5
ICI 1 & 2
PGM 3
09/19/2022 Causal Models
Conditional distributions are not enough to model and estimate causal effects. In this lecture, we'll see how the notion of interventions is a general way to quantify causal effects. In particular, we'll use causal graphs and structural causal models (SCMs) to formally define interventional distributions.
ECI 2, 6.1-6.3
ICI 3, 4.1-4.3
C 1.2-1.4
Inverse CDF: link1, link2
09/26/2022 Identification
In order to estimate causal effects from data, we first need to convert them to a function of data (observational distribution). This process is known as identification. We use causal graphs to study sufficient conditions for identification, such as Back-door and Front-door criteria. We then introduce do-calculus as a complete non-parametric identification algorithm.
C 3.1-3.4
ECI 6.4, 7.1-7.2
ICI 6.1-6.2
10/03/2022 Estimation
Once we have identified a causal query, we still need to build efficient estimators to estimate it from finite data. This lecture focuses on some well-known estimators in the literature, including parametric g-formula, propensity scores, inverse propensity weighting, and matching.
ICI 7.1-7.6
TAR-Net
Propensity Score
10/10/2022No Class - Thanksgiving -
10/17/2022 Instrumental Variables + Student Paper Presentation ICI 9.1-9.4
ECI 9.3, link
10/24/2022Student Paper Presentation -
10/31/2022Student Paper Presentation -
11/07/2022No Class - Reading Week -
11/14/2022 Invariant Learning + Student Paper Presentation ICP paper
11/21/2022 Double Machine Learning DML paper
DML application
11/28/2022Student Project Presentation -
12/05/2022Student Project Presentation -