Datawarehousing Terms
Datawarehousing Terms#
- Data mining - showing unknown trends
-
If insights are not correlated they are useless
-
BI - reporting, analytics, management information, based on facts
- BI - Gain insights and make trusted decisions
- Advanced analytics - extracts fact
- Operational data store / data lake - a staging area
- Data warehouse purpose - enhance rate of knowledge acquisition and answering management questions
Can’t do it on an operational system - must be done on a data warehouse.
Don’t want to disturb operational (transactional) data
Can’t do both. Operations must be independent.
Quick wins with datascience scripts after a while it will use more resources.
- Components of a data warehouse?
- Managing relationship and syncing of operational database to Datawarehousing - data duplication.
Single source of truth
KPI:
- Cost of Capital
- Headcount
- Earnings per share
- Cost per Unit
Users in BI:
- Strategic - Executive make more decisions
- Tactical - Data scienctists, make some decisions.
-
Operational - Little to no decision.
-
Data warehouse Enterprise Information Management Framework - Data sources, Integration, Derived Information, Semantics (Presentation Logic), Deliver / Manipulation / Consumption
-
Cube is a precalculated values stored in memory
- Star schema: A fact and dimensions around it
- Fact: Quantity, cost, total
- Dimensions: Time, Customer, Product, Branch
Delta processing
Surrogate key is an extra key for delta processing and seeing changed records