Table of contents
No. ODAP is a use case framework which consists of three parts:
ODAP Framework | Databricks Feature store |
---|---|
Simple, well defined interface for writing features | General interface for saving DataFrames |
Optimized for reading performance | User is responsible for optimization |
Optimized for writing multiple features at once | User is responsible for optimization |
Optimized for writing features with time windows | User is responsible for optimization |
Special interface for generating target based training sets | User is responsible for generating targets |
Fully customizable per feature metadata | Only models/features/notebook links as metadata |
Outsources online serving to Databricks FS | Handles online serving out of the box |
For more in depth explanation of ODAP advantages read Feature factory overview
.py
files in feature notebooks?