Individual Expected Completion using Logistic Generalized Additive Mixed Models

Abstract

Michael Lopez posted a great article not long ago explaining how Generalized Additive Models (GAMs) are an excellent way to measure the non-linear effects of explanatory variables x on response variable y. Lately, I’ve been playing around with linear and logistic mixed-effects models, so I thought about combining these with GAMs to estimate the probability of completion per Quarterback while accounting for non-linearities, especially on air yards.

Adrian Cadena-Medina

Data Analyst Advisor

Driving meaningful improvement by employing data science. Statistically oriented with interest in Machine and Deep Learning. My research interests include sports analytics, sustainability, and welfare economics.