Data Warehouse and Data Mining
One book, placed in its field — what Jugnesh Kumar's book gets right, where it goes further than the rest, and what the rest of the field adds.
Mike West, Editor · Leading With Data · dataengineerprofessional.com/guides/profile-data_warehouse_and_data_mining_concepts
Arrow keys or swipe to move · N for notes · F for full screen
Data Warehouse and Data Mining
One book, placed in its field — what Jugnesh Kumar's book gets right, where it goes further than the rest, and what the rest of the field adds.
Mike West, Editor · Leading With Data · dataengineerprofessional.com/guides/profile-data_warehouse_and_data_mining_concepts
The problem
Decision-Making Quality erodes when it is left to instinct instead of method.
You were taught the moves piecemeal, never the whole model.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
The model
- 01Data Warehouse Architecture Design
- 02Schema Design Choice
- 03ETL Process Quality
- 04Metadata Management
- 05Data Mining Technique Selection
- 06Data Quality
- 07Query Performance
- 08Analytical and Knowledge Discovery Capability
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Step 1 of 5 · Data Warehouse Architecture Design
Use three-tier when load and query workloads genuinely compete for resources; otherwise the extra layer is overhead.
Misconception
Practitioners assume three-tier is always the 'enterprise-grade' right answer and single-tier is amateur.
Reality
The tier count is a decoupling decision, not a quality ranking; the three-tier separation of staging, reconciled store, and presentation buys you isolation between load and query workloads, which only pays off when those workloads actually contend.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Step 2 of 5 · Schema Design Choice
Star schemas win for query performance and BI-tool compatibility in the vast majority of cases.
Misconception
Teams believe snowflaking dimensions is 'more normalized and therefore more correct' and thus the disciplined choice.
Reality
Normalization is a transactional virtue that becomes a liability in analytics; snowflaking trades query speed for storage savings that are trivial at warehouse prices, so its extra joins usually cost more than they save.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Step 3 of 5 · ETL Process Quality
Validate data quality inside the ETL, before it reaches the warehouse — post-load fixes are far more expensive.
Misconception
Practitioners treat ETL as a one-time build-and-forget plumbing task once the pipeline 'runs green.'
Reality
A pipeline that completes successfully can still load wrong data; source schemas drift, business rules change, and reference data goes stale, so ETL quality is a continuous validation discipline, not a build milestone.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Step 4 of 5 · Metadata Management
Automated lineage lets you assess the blast radius of any source or transform change instantly.
Misconception
Metadata is seen as documentation overhead you produce for auditors, separate from the real work of moving data.
Reality
Metadata is the operational nervous system: lineage tells you what breaks when a source changes, and shared definitions are what let two teams agree that 'active customer' means the same thing — it directly drives both quality and analytical trust.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Step 5 of 5 · Data Mining Technique Selection
Frame the business question and data structure before choosing an algorithm.
Misconception
Teams pick the most sophisticated or fashionable algorithm, assuming complexity yields better insight.
Reality
Technique selection is dictated by the problem shape and the data, not by algorithmic prestige; a well-framed association-rule mining can deliver more business value than a deep model applied to the wrong question.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
A framework to use
Three-Tier Data Warehouse Architecture
This framework is part of the full guide. See the guide →
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
The cheat sheet
- Data Warehouse Architecture DesignUse three-tier when load and query workloads genuinely compete for resources; otherwise the extra layer is overhead.
- Schema Design ChoiceStar schemas win for query performance and BI-tool compatibility in the vast majority of cases.
- ETL Process QualityValidate data quality inside the ETL, before it reaches the warehouse — post-load fixes are far more expensive.
- Metadata ManagementAutomated lineage lets you assess the blast radius of any source or transform change instantly.
- Data Mining Technique SelectionFrame the business question and data structure before choosing an algorithm.
- Data QualityMeasure all four dimensions — accuracy, completeness, consistency, timeliness — not just accuracy.
- Query PerformanceMatch partition keys and indexes to real query filters, not to intuition.
- Analytical and Knowledge Discovery CapabilityAnalytical capability is bottlenecked by its weakest upstream input, not its fanciest tool.
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data
Sources
- Data Warehouse and Data Mining — Jugnesh Kumar
Data Warehouse and Data Mining · Mike West, Editor · Leading With Data