Data mining

Data mining searches large data sets for patterns people would never spot manually — powering recommendations, fraud alerts and forecasts.

• Techniques: classification (assign categories), clustering (find natural groups), association rules ("bought bread → often buys milk"), anomaly detection, prediction.
• Needs clean, integrated data — usually mined from a data warehouse.
• Business uses: market-basket analysis, churn prediction, credit scoring, medical research.
• Ethical concerns: privacy, profiling, discrimination from biased data, lack of transparency.
• Correlation is not causation — mined patterns require human interpretation.

• Classification vs clustering — sorting into KNOWN categories vs discovering UNKNOWN groups.
• Anomaly detection — flagging unusual records (e.g. fraudulent card use).