Encoding Multi-Layered Vega-Lite COVID-19 Geodata Visualizations with MLflow

Explore the process of creating and managing multi-layered Vega-Lite visualizations using COVID-19 and Geodata in this 54-minute presentation from Databricks. Learn how to leverage Vega-Lite for encoding visualizations as JSON objects and utilize the MLflow model registry as a visualization registry. Follow along as the presenters demonstrate the creation of complex visualizations, including base visualizations, geo visualizations, and multiple visualization styles. Discover techniques for data manipulation, visualization testing, and editing. Gain insights into using helper functions, registering models, and searching for specifications. The presentation includes a Q&A session and a demo showcasing code snippets, visualization styles, and target runtimes. Access the accompanying GitHub repository and interactive demo for hands-on practice.

Introduction
Andy Bauman
James Hibbard
What are models
What are visual models
Launch MLflow server
MLflow UI
VegaLite
P Sofa
Visualization
Copy Paste
Remove Data
Repopulate Data
Testing Visualization
Editing Visualizations
Fake Data
Base Visualization
Helper Function
Register Model
Search Model
Search Specification
Geo Visualization
Spec
Geojson
Cases
Filtering
Selecting by State
Register visualizations
Search for statuses
Rerun search
QA session
Demo
Style
Code snippet
Visualization styles
Multiple visualization styles
Target runtimes
QA