What Does It Take to Be a Good Fisherman?

Explore the challenges and complexities of learning from strategically-influenced data in this thought-provoking lecture. Delve into the concept of self-selection bias and its impact on statistical models, using the analogy of fishermen to illustrate how data collection can miss crucial information. Examine recent progress in addressing econometric challenges, including estimating linear models under self-selection bias and identifying non-parametric auction models. Discover the implications of this research for various fields, from understanding student and employee performance to learning from expert demonstrations and analyzing strategic behavior in markets. Gain insights into potential solutions and open directions for future investigation in this critical area of study.

Introduction
Statistical Approach
SelfSelection Bias
Imitation Learning
Selective Reporting
Market Disequilibrium
Nonparametric Setting
Summary
Applications
Linear Models
Comparison
Assumptions
Selector Function
Results
Lucky Problem
Gradient Descent
Conclusion