NFL Analytics Suite

R
xgboost
tidymodels
Models and visualizations of fourth-down decisions, rushing and passing yards over expected, and team performance across four NFL seasons.
Published

February 1, 2026

Overview

Built logistic and linear regression models in R to predict 4th-down go decisions and quantify Air/Rush Yards Over Expected (AYOE/RYOE) across 4 NFL seasons, controlling for game situation and opponent quality. Trained an XGBoost model ranking 80+ NFL backs by rush yards over expectation and produced publication-quality visuals covering EPA/play, pass rate, and field goal metrics league-wide.

Visuals

4th Down Go/No-Go Model (2021–25)

Air Yards Over Expected Model (2021–25)

EPA/Play by Team

Run/Pass Rate by Player — Bears 2025

EPA per Play vs. Pass Rate (2022–25)

Field Goal Rate: Field vs. Result vs. Grade (FG_OE)

Pass Rush EPA (2025)

Quarterback Grade

Rush Yards Over Expected — XGBoost Rankings

Tech Stack

R, ggplot2, nflreadr, xgboost, tidymodels, dplyr