epswartz / scalable-league-win-prediction

Scaling a basic ML model (predicting League of Legends wins) in GKE.

scalable-league-win-prediction

A scalable service for predicting wins in League of Legends using logistic regression. Built to demonstrate containerized scalability on GKE.

Running Locally

1. python3 main.py
2. To test with httpie: http POST localhost:8080/predict < test_payload.json

Deployment on GKE

1. Create a GKE cluster
2. kubectl apply -f ./kube/deployment.yaml
3. kubectl apply -f ./kube/service.yaml

Load Testing

1. pip install locust
2. make locust host=<HOST>