Design the data model and detailed database schema to store model training data in the database.
Success metrics
Goal
Metric
We should be able to ingest retrospective data (that is customer retrospective data) into a database.
Successful ingestion of data.
Assumptions
Requirements
Requirement
Importance
Notes
Data model should be designed so that data scientists are able to train, test and validate ML models
HIGH
Every customers data set should be logically separate. Does this data model accomplish that?
HIGH
User interaction and design
Open Questions
Question
Answer
Date Answered
Sally Hong When we go live, we will need to ingest and store the data that we will receive via APIs. Will that data be saved in the same data model or do we need a different data model for that?
- The Data Model (Generic) should be the same across all hospitals - We can make slight tweaks depending on the production model - The model takes into account raw data (CV ID’s and demographic data) separated and transformed data (post-cleaning where we denormalized data - Denormalized = consolidated data for data analysis / ML