There code only requires the standard installation of Anaconda Python. It will requires Pip to install the pandas, numpy, sklearn, and xgboost libraries.
Important Note: The XGBoost version used for this project is 0.81. Using version 0.90 will result in different results.
The purpose of this project is to continue improvement one of my projects to incorporate lessons learned and additional techinques.
There are two folders. One for data analysis and one for machine learning. Each folder contains a copy of the data.
The data analysis folder contains 2 python scripts. The first one is for supporting functions.
The machine learning folder contains 4 python scripts. The first one is for supporting functions and there is one more parameter tuning, feature selection, and building the final model. This is done primarily for ease of use.
The parameter tuning and feature selection results are available in “Results” folder.
The final report is available as a PDF within the GitHub respository below.
The GitHub repository in located here: here
There will be weekly Medium posts tracking the performance on the model for the NFL 2019 season.
Week 2
Week 3
Week 4
Week 5
Week 6
Week 7
Week 8
Week 9
Week 10
Week 11
Week 12
Week 13
Week 15
NFLDB – https://github.com/BurntSushi/nfldb
NFL Betting Data – http://www.footballlocks.com/nfl_lines.shtml
NFL Weather Data – http://www.nflweather.com/
https://github.com/dkim319/NFL_Predictive_Model_v2
Leave a Reply