48113 - Decision-Making Support Information Systems

Learning outcomes

The objective of the course is to describe and explain data warehousing technologies and the process of knowledge discovery in databases, and their role in business intelligence. In particular, prominent data mining algorithms are described, and their operational features and application prerequisites are analysed. The differences between computational requirements in centralized, distributed, and stream data environments and the basic techniques to satisfy them are discussed. The course includes laboratory classes during which implementations of the most popular data mining algorithms are experimented with on real data sets.

Course contents

Data warehousing. Data warehousing and business intelligence. OLTP e OLAP. Architecture of a data warehouse. Schemata and operations in a data warehouse.
Knowledge Discovery. Knowledge discovery and business intelligence. The knowledge discovery process. Data Mining Algorithms: Association rules: The APRIORI algorithm; clustering vector data: One pass algorithms, the BIRCH algorithm; density-based clustering algorithms: The DBSCAN algorithm, the DENCLUE algorithm; clustering categorical data: outline; clustering metric data: outline. Algorithms for Distributed Data Mining: Privacy and cooperation in mining distributed data; distributed algorithms for association rules; distributed algorithms for clustering; distributed algorithms for classification. Algorithms for mining data streams: Computation of aggregates on stream data; clustering data streams: Pyramidal time frames and the CluStream algorithm. Laboratory classes: IBM Intelligent Miner, Microsoft SQL Server.

Readings/Bibliography

Jiawei Han and Micheline Kamber, Data Mining: Concepts and Techniques, The Morgan Kaufmann Series in Data Management Systems, Jim Gray, Series Editor. Morgan Kaufmann Publishers, August 2000. 550 pages. ISBN 1-55860-489-8

Teaching methods

Operation, limits of applicability and computational complexity of the most prominent data mining algorithms are explained during frontal lessons. During laboratory classes, commercial software tools are experimented with on real and artificial data sets and the results are collectively discussed.

Assessment methods

• Practical examination with data warehousing and data mining tools
  
• Oral examination

Teaching tools

• PC and overhead projector
• Laboratory with desktop PCs
Software:
• Database Management System IBM DB2 Express-C
• IBM Intelligent Miner
• Microsoft SQL server
  

Links to further information

http://www-db.deis.unibo.it/~slodi/SISD/2008-2009/sisd.html

Office hours

See the website of Stefano Lodi