doitintl / vizier-presentation

Sample Workbooks for Vizier

• This repo is accessible at http://bit.ly/viziergithub
• See the presentation at https://bit.ly/vizierpresentation

Workbooks

Sample for GAPIC Vertex API

• /gapic-vizier-fraud.ipynb
• Uses the Vertex AI Vizier API. Generally available.
• This notebook is the only avialable sample for Machine Learning with Vertex AI Vizier, after the ML sample mentioned below.
• This is the recommended API as the Vertex platform is Google's main direction.
• Step through by the guidepoints tagged #[0], #[1], #[2], etc.

Sample for Discovery API

• /discovery-api-fraud.ipynb
• Uses the discovery-based API as part of Google AI Platform. Beta as of 11/2022.
• This is the recommended API as the Vertex platform is Google's main direction.
• Step through by the guidepoints tagged #[0], #[1], #[2], etc.

For you to try: A Notebook based on Ryan Holbrook's Major League Baseball Notebook

• /mlb/vertex-ai-with-mlb-player-digital-engagement.ipynb
• From this on Kaggle https://bit.ly/viziermlb
• Improvements here:
  • Shows my run output
  • Auto-sets project
  • Allows non-home-dir run
  • Delete import (which is not actually used by the code ) that triggers error
• How to run that Notebook
  • Read the NB itself for instructions
  • For the NB to auto-download the data, you need a kaggle.json auth key from Kaggle.
  • You can mostly run it end-to-end, but see my minor fixes.