Targeted Learning

Targeted Learning

Instructors: Mark Van der Laan, Ph.D.

Image credit:
Keegan Houser

Course Summary

This course teaches students to construct efficient estimators & obtain robust inference for parameters that utilize data-adaptive estimation strategies (i.e., machine learning). Students perform hands-on implementation of novel estimators using high-dimensional data structures, providing students with a toolbox for analyzing complex longitudinal, observational & randomized control trial data. Students learn & apply the core principles of the Targeted Learning methodology, which generalizes machine learning to any estimand of interest; obtains an optimal estimator of the given estimand, grounded in theory; integrates state-of-the-art ensemble machine learning techniques; & provides formal statistical inference in confidence intervals & testing.

Overview of Lectures and Related Notes

Part I

Lecture 1 & Notes 1: Roadmap of Statistical Learning

Lecture 2 & Notes 2: Examples of Data Generating Experiments

Lecture 3 & Notes 3: Traditional Data Analysis

Lecture 4 & Notes 4: Structural Causal Model (SCM), Causal Quantities, Identification

Lecture 5 & Notes 5: Interventions

Lecture 6 & Notes 6: Understanding Nonparametric Density Estimation - Super Learning of a Density 

Lecture 7 & Notes 7: Super Learning and Oracle Inequality, Optimal Individualized Treatment and Prediction of Survival

Lecture 8 & Notes 8: Super Learning of a Conditional Density or Conditional Multinomial Distribution

Lecture 9 & Notes 9: Integrals with respect to Measures

Lecture 10 & Notes 10: Highly Adaptive Lasso (HAL)

Part II

Lecture 11 & Notes 11: Online Super Learning

Lecture 12 & Notes 12 : Empirical Probability Measure

Lecture 13 & Notes 13: Functional CLT for Empirical Processes

Lecture 14 & Notes 14: Asymptotic Linearity of an Estimator

Lecture 15 & Notes 15: Functional Delta Method

Lecture 16 & Notes 16: Canonical Gradient/Efficient Influence Curve

Lecture 17 & Notes 17: Tools for Computing Projections and Canonical Gradient

Lecture 18 & Notes 18: Efficiency Theory

Lecture 19 & Notes 19: Efficiency of NPMLE

Lecture 20 & Notes 20: One-Step Estimation

Lecture 21 & Notes 21: TMLE

Readings

The following textbook is a helpful resource when diving into this course. 

-  Please find the resource available here. 

Thank You!

This work by CTML Faculty and/or Staff is licensed under CC BY 4.0