Overview
A Flask service serving an XGBoost model from model/xgboost_model.json, which train.py produces if it is missing. It exposes a health endpoint and a predict endpoint that takes a one or two dimensional feature array and returns the prediction with its input shape, answering 400 on invalid input. Docker and Kubernetes manifests are included.
Installation
git clone https://github.com/coccinella-labs/xgboostapi