Speed Ratings

The goal of this project is to create an accurate machine learning model to predict the speed rating of all finishers in any given race in New York State. The model will be trained, valuated, and tested on speed ratings from 2014 to 2019 along with weather data from the National Centers for Environmental Information and course specific information from Tully Runner's and Milesplit. The hope is that this model can be applied to future high school cross country races and produce a higher quantity of accurate speed ratings.

The repository is organized as follows:

data/ contains all of the raw and preprocessed data used in the project
figures/ contains all of the figures generated in the project
report/ contains reports on the development and results of the project
results/ contains all of the results generated in the project (including the final trained models)
src/ contains all the code. The code is organized as follows:
- compile.ipynb contains code for compiling the raw data
- preprocessing.ipynb contains code for preprocessing the compiled raw data
- eda.ipynb contains code for exploratory data analysis
- elasticnet_ml.ipynb contains code for training and testing the elastic net model
- randomforest_ml.ipynb contains code for training and testing the random forest model
- xgb_ml.ipynb contains code for training and testing the xgboost model
- knn_ml.ipynb contains code for training and testing the k-nearest neighbors model
- evaluation.ipynb contains code for evaluating the final models

This project is built in Python 3.11.4 and uses the following key packages:

joblib 1.3.2
matplotlib 3.7.2
numpy 1.24.4
pandas 2.0.3
scikit-learn 1.3.0
scipy 1.11.3
seaborn 0.12.2
shap 0.42.1
xgboost 1.7.6

The environment.yml file contains a full list of the packages for this project.

johnsfarrell/speed-ratings

Speed Ratings