Boxing Fighter Pay - Multiple Regression

Using boxing fighter pay data to introduce multiple regression, RMSE, and residual plots in Python

Author

Affiliation

Izaan Khudadad

University of North Carolina at Charlotte

NoteFacilitation notes

This module was originally developed as a Jupyter notebook. The SCORE preprint server does not run the notebook directly, so users should download the materials below and run the notebook locally or in a Jupyter-compatible environment.

Students should be provided with the following files:

• Student-facing Jupyter notebook
• Boxing fighter pay data

Additional instructor materials are available below:

• Solutions Jupyter notebook

Introduction

In the following activity, you will use data compiled by Peter Anderson from the Nevada State Athletic Commission(NSAC) and the California State Athletic Commission(CSAC). The dataset covers professional boxing fights held between 2009 and 2017, which includes individual fighters, their opponents, fight characteristics, broadcasting networks, and fighter earnings.

Each observation represents a single fighter in a given bout, forming a structure that enables tracking of fighters over time.

Using this data, you will work on how various variables are correlated to fighter purse amount, as well as learn about how to determine the quality of a regression model.

Learning Goals

By the end of the activity you should be able to:

1. Use Python to create models based on more than two predictors.
2. Understand what Root Mean Squared Error(RMSE) is in the context of multiple regression.
3. Look at Residual plots to determine the quality of linear regression models.
4. Use a variety of Python Libraries such as MatplotLib, Numpy, and Seaborn in creating models.

Data

The dataset includes over 4,600 fight entries and more than 1,200 unique professional boxers. Each row represents one fighter in one fight.

• Boxing fighter pay data

Variable Descriptions

Variable	Description
Boxer	Name of the boxer (Last, First)
Date	Date of the fight (YYYY.MM.DD)
Purse	Reported purse (fighter’s earnings) in USD
lnRPurse	Natural logarithm of the purse (for regression use)
weight	Weight of the boxer (in pounds)
Age	Age of the boxer at the time of the fight
Wins	Number of professional wins prior to the fight
Losses	Number of professional losses prior to the fight
KO	Number of professional knockout wins prior to the fight
W-Title	Indicator for world title bout (1 = yes, 0 = no)
PPV	Fight was on Pay-Per-View (1 = yes, 0 = no)
RDS	Scheduled number of rounds

Module Files

The full module materials are linked below.

• Student-facing Jupyter notebook
• Solutions Jupyter notebook
• Boxing fighter pay data

Supporting Images

The module also includes the following residual plot images:

• Residual plot A
• Residual plot B
• Residual plot C
• Residual plot D

Module

This shell page provides access to the module materials. Download the student-facing Jupyter notebook and the supporting data file above to complete the activity.

To run the notebook, users will need a Python environment with packages such as pandas, numpy, matplotlib, seaborn, statsmodels, and scikit-learn.

Summary

In the provided material, students explored how to use multiple linear regression to model Boxing data. By building, evaluating, and interpreting regression models, students practiced core data science skills such as:

• Selecting and preparing predictor variables
• Understanding how each predictor influences the response
• Evaluating model performance using RMSE
• Diagnosing fit using residual plots
• Reading and interpreting a full regression summary