Dockerized Machine Learning Rest API

---

Example of turning an XGBoost model (built here) using Flask, Docker and Google Cloud Run.

Local Use

---

Build the image

docker build -t mlapp app

Run the API locally:

docker run -ip 8080:8080 mlapp

From Python:

data = {'columns': ['Energy_Star', 'Site_EUI', 'NGI', 'EI', 'Residential'],
      'data': [[7744, 52.7, 45.9641802469, 10.1204555556, 1],
               [9, 112.2, 25.9187489356, 29.7707095517, 1]]}


import requests

result = requests.post(url="http://0.0.0.0:8080/predict",
                     json=data)

result.json() # [0.0031427741050720215, 0.007205158472061157]

Deploying to Google Cloud Run:

---

Build image on Google Cloud:

cd app

gcloud builds submit --tag gcr.io/<project-id>/mlapp

Deploy to Google Cloud Run:

gcloud run deploy --image gcr.io/<project-id>/mlapp --platform managed