Clustering on NBA League From 2003 to 2018
Project Overview
Using unsupervised machine learning to analyze player decline in the NBA based on the player archetype. This analysis will provide assistance to NBA teams for roster building and player adjustments.
The analysis will aim to provide insight on the following questions:
- Determining the new player archytpe in the modern day position list.
- Will adjusting the athletes playstyle help improve performance.
Datasets
- Kaggle Player Data
- A data sample has been drafted.
Environment
cd db && python3 -m venv env
source env/bin/activate
pip install -r requirements.txt
cd db && export PYTHONPATH=$PWD
Additional packages
cd db
pip freeze > requirements.txt
FastAPI - Backend
cd db && source env/bin/activate
cd db && export PYTHONPATH=$PWD
cd db && python app/main.py
Run /transformLoad
Run /ml/pca
Run /ml/timeseries
Make sure Mongo is running
Frontend - Plotly Dash
cd db && source env/bin/activate
cd client && python app.py
Upload the file "TO_UPLOAD_FULL_SeasonsDataRaw" for a full test.
MongoDB
mongo
show dbs
use players
db.Cleaned_Dataset.find()
Team Project - UofT Bootcamp
Communication Protocols
- Slack: Team discussions, questions, suggestisons & resource sharing.
- Google Meet: Team meetings, discussions
- Trello: Work organization, scheduling