TL;DR

A few news about Orbit:

  • a new blog post about the v1.1
  • the project RADAR – collaboration with Fraud team in applying Orbit on anomaly detection
  • a recently developed online dash of orbit-ml PyPI downloads statistics

The Dashboard

There is a lot of fun in learning the new technology in building automatic dashboard with Python. The core components of the flow are

  1. credential modules - building credential from Github secret keys to .json files.

  2. BigQuery modules - using credentials set to download statistics

  3. offline data message - a few utils to clean and packing data from raw to derive and then to clean as one final data source.

  4. data pipeline and automation - setting up the trigger and cadence of tasks with Github Actions

  5. dashboard building and automation - using python package plotly and dash to prototype dashboard and deploying thru Heroku

After all these learning, spent an hour or so to have fun with picking colors and css stuff :)