Skip to main content

Data warehouse and data mining

Short answer

A data warehouse is an integrated, subject-oriented, time-variant, non-volatile repository for analysis, fed by ETL from operational systems. It avoids slowing operational DBs, consolidates historical data, and provides clean input for data mining, which discovers patterns like associations and clusters.

What is a Data Warehouse? Why do you require data warehouse along with operational databases? How is it related with data mining? (Discuss with a suitable diagram.)

Data warehouse

A data warehouse is a central repository of integrated, subject-oriented, time-variant and non-volatile data collected from many sources, kept for analysis and decision support (Bill Inmon's definition).

  • Subject-oriented - organised by subjects such as customer, sales.
  • Integrated - consistent formats from many sources.
  • Time-variant - stores historical data.
  • Non-volatile - data is not changed after loading.
 Operational DBs --\
 Files/ERP/CRM ----- ETL --> Data Warehouse --> OLAP, reports
 External data ----/             |                 Data mining

Why a data warehouse is needed with operational databases

  1. Operational databases are tuned for fast transactions; heavy analytical queries would slow them.
  2. They hold only current data, while analysis needs years of history.
  3. Data is scattered over many incompatible systems; the warehouse integrates and cleans it.
  4. Operational data is detailed and normalised; the warehouse is designed (star schema) for fast aggregation.
  5. It gives a single consistent version of data for management and avoids interference with live operations.

Relation with data mining

Data mining discovers hidden patterns, trends and relationships (classification, clustering, association rules) in large data sets. The warehouse supplies clean, integrated, historical data, which is the ideal input for mining. Mining results (customer segments, fraud patterns, market baskets) are used in decisions and may be stored back in the warehouse.

 Warehouse -> Select/Preprocess -> Mining -> Patterns -> Decisions

✓ Answer checked by Claude Opus 5.5 (high effort)

Related questions

All Information System chapter 4 questions →