COURSES

Goals

Knowledge discovery in databases is the process of discovering patterns and models, described by rules or other human-understandable representation formalisms. The most important step in this process is data mining, performed by using methods, techniques and tools for automated constructions of patterns and models from data. The course objectives are to (a) introduce the basics of data mining, the process of knowledge discovery in databases, and the CRISP-DM methodology, (b) present selected data mining methods and techniques, and (c) present the methodology for result evaluation. The students will master the basics of data preprocessing, data mining, and knowledge discovery and will be capable of using selected data mining tools and results evaluation methods in practice.

Curriculum

Introduction: introduction to data mining and knowledge discovery in databases, relation with machine learning, visualization of data and models, presentation of the CRISP-DM knowledge discovery methodology. Data mining techniques: presentation of specific data mining techniques: decision, regression and model tree learning, learning classification and association rules, clustering, nearest neighbors approach, Naive Bayesian classifier, support vector machines, artificial neural networks, subgroup discovery, ensemble classifiers. Heuristics and results evaluation: presentation of search heuristics, heuristics for estimating the quality of induced patterns and models, methodology for results evaluation. Advanced data mining methods: semi-supervised learning, active learning, relational data mining, propositionalization, semantic data mining. Practical training: practical use of selected data mining techniques and tools.