Value-Driven Analytics Development
summary
Argues that prioritizing business value shortens time to insight by prototyping on a lake before heavy modeling, showing examples: land raw clickstream in lake and run ad hoc queries from production data with prototype with PySpark notebooks. Uses metadata-driven cleansing and automated validation to avoid deploying staging tables and to cut developer lead time; cites storage in Azure lakes as roughly one-fifth the cost of databases. Productionize only after value is proven, or skip expensive modeling if ROI is negative.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:40Simon Whiteley
- 1:32What are we doing
- 3:46The big problem
- 4:50The traditional way
- 11:00Data lakes
- 13:32Warehouses
- 16:28Modern Data Warehouse
- 19:51ValueRealization
- 20:59Data Ops
related talks
30mValue Engineering: How to Get Your Big Data Investment.....
47mData Analytics: Realizing the Full Potential
54mBehind the Hype - Architecture Trends in Data
43mData Lake beyond the Buzzword
24mA Practical Approach To Unlock Value From Data and Analytics
18mBehavioral Data Creation for AI - Delivering Value and Insights
22mData as a Product: Transforming Analytics for Business Success
21mPutting big data self-service analytics in the hands of business people