Data Mining

Finding hidden information in a database; fit data to a model.

	Database	Data Mining
Query	Well defined; SQL	Poorly defined; no precise query language
Data	Operational data	Not operational data
Output	Precise; subset of database	Fuzzy; not a subset of database

Classification: map data into predefined classes
- Supervised learning
- Pattern recognition
- Prediction
Regression: map data to real valued prediction variable
Clustering: groups similar data together
- Unsupervised learning
- Segmentation
- Partitioning
Summarization: map data into subsets with associated simple descriptions
- Characterization
- Generalization
Link analysis: uncover relationships among data
- Affinity analysis
- Association rules
- Sequential analysis
Time series analysis
- Future value prediction
- Pattern identification
- Behavior classification

Metrics
- Usefulness of results
- Return on investment (ROI)
- Accuracy of results
- Space/time required
Social implications
- Privacy
- Profiling
- Unauthorized use
Database perspective
- Scalability (large data)
- Real world data (noisy data)
- Updates (dynamic data)
- Ease of use
Visualization techniques
- Graphical
- Geometric
- Icon-based
- Pixel-based
- Hierarchical
- Hybrid

Knowledge Discovery in Databases (KDD)

Process of finding useful information & patterns in data. Data mining uses algorithms to extract information & patterns derived by the KDD process.