MLHL

Attempt at using ML and DL to perform NHL fantasy drafts for the 2024-2025 Season

Data Sources

Skater Data - NHL.com

Goalie Data - NHL.com

Setup

Development was performed on Fedora 40 with no dedicated GPU. Adapt the below commands to your preferred platform.

Conda Environment setup

conda create -n mlhl-env python=3.12
conda activate mlhl-env

Scraping Dynamic Data

Scraping is unnecessary unless you want to collect your own dataset. CSV files are available and scraping can be skipped if desired.

conda install selenium pandas
python -m pip install webdriver-manager
python -m pip install --upgrade selenium

Pytorch Forecasting

conda install pytorch torchvision torchaudio cpuonly -c pytorch
python -m pip install pytorch-forecasting
conda install tensorboard

Scikit-Learn Forecasting

conda install matplotlib scikit-learn

You can optinally install Jupyter to visualize the data (.ipynb to be uploaded shortly)

conda install jupyter