What Comes After The Star Schema? Dimensional Modeling For Enterprise Data Hubs
14 likes8,102 views
The document discusses advancements in dimensional modeling for enterprise data hubs, emphasizing the limitations of traditional star schemas and introducing the 'supernova' model as a solution for analytics. It outlines the design rules for creating a supernova schema, which involves identifying root dimensions and relevant facts while de-normalizing critical child dimensions. The content highlights the importance of operationalizing insights and evolving data science practices for improved analytics performance.
What Comes After The Star Schema? Dimensional Modeling For Enterprise Data Hubs
- 1. 1 What Comes After the Star Schema? Dimensional Modeling for Enterprise Data Hubs Josh Wills Senior Director of Data Science
- 2. About Me 2
- 3. Two Fish 3
- 4. 4 Once Upon A Time…
- 5. Our data looked like this… 5
- 6. …but the data analyst wanted this. 6
- 8. The Unicorn Solution 8
- 9. The Great Compromise 9
- 10. Ease of Use 10
- 11. Division of Labor 11
- 12. Given Away Freely 12
- 13. The Central Idea of an Entire Industry 13
- 14. Modern BI and Reporting 14
- 15. 15 The Next Hard Problem
- 17. The Analytics Maturity Curve 17
- 18. Example: Modeling Risk 18
- 19. The Impedance Mismatch, Redux 19
- 20. The Unicorn Solution, Redux 20
- 21. This All Feels Very Familiar 21
- 22. It’s Time To Find A New Compromise 22
- 23. Data Science 101 23
- 25. A Simple Star Schema for Search 25
- 26. The Cartesian Explosion 26
- 27. A Supernova Schema for Search 27
- 28. Beyond Analytic SQL Functions: Nested SQL Sessions 28
- 29. Event Series Analytics 29
- 31. 31 Supernova for BI and Reporting
- 32. A New Kind of Data Model 32 Context/Backend RDBMS NoSQL Operational Entity-Relational Generic Objects Analytical Star Supernova
- 33. From Star To Supernova 33
- 34. Supernova 101: Affinity Analysis 34
- 35. Supernova Design Rules 1. Identify the root dimension. 2. Identify the facts about the root dimension that we want to analyze. 3. De-normalize critical child dimensions. 4. Define the time window. 35
- 36. Creating a Supernova 36
- 37. Injecting Child Dimensions 37
- 38. Taking Advantage of Multi-Insert 38
- 39. 39 Supernova for Operational Analytics
- 40. A New Kind of Data Model 40 Context/Backend RDBMS NoSQL Operational Entity-Relational Generic Objects Analytical Star Supernova
- 41. Data Science and the Holy Grail 41
- 42. Replicating the Online Environment 42
- 43. Feature Engineering: Within / Across 1. Generate normalization and segmentation features. 2. Generate normalization constants and segments using the result of Step 1. 3. Generate input features using the original data and the result of Step 2. 4. Generate a model using the result of Step 3. 43
- 44. Integrating New Signals 44
- 45. Building a Wall 45
- 47. Thank you! Josh Wills, Senior Director of Data Science, Cloudera @josh_wills