Instructor: Ratko Orlandic, email: rorla2@uis.edu

Course Description: This course teaches advanced techniques for discovering hidden patterns in the rapidly growing data generated by businesses, science, web, and other sources. Focus is on the key tasks of data mining, including data preparation, classification, clustering, association rule mining, and evaluation.

Course Objectives: Because the development of data mining systems requires highly skilled programmers/ problem-solvers, data mining is one of the highest paid professions in information technology. The requirements of this profession include: ability to conceptualize the problem at hand; ability to investigate/research the problem; ability to select methods and techniques appropriate for the task; and the ability to develop the methods and tools for the given task. This, in turn, requires understanding of data-mining problems and techniques, excellent programming skills, and virtually permanent self-development. In the light of this, the course objectives are organized to help students: understand the fundamental processes, concepts and techniques of data mining and develop an appreciation for the inherent complexity of the data-mining task; advance relevant programming skills; and advance research skills through the investigation of data-mining literature.

Outline of Topics to be Covered:

• Introduction (definition, motivation, applications, tasks);
• Basic concepts (concept, instance, attribute, types of learning, data warehousing and OLAP);
• Data preparation (data cleaning, data integration, data transformation, data reduction, discretization, concept-hierarchy generation);
• Concept description (data summarization, attribute oriented induction, attribute relevance analysis);
• Classification and numeric prediction (1R, Bayesian classification, decision trees, instance-based classification, support vector machines, regression);
• Evaluation (evaluation steps, balancing, cross-validation, stratification, bootstrapping, evaluation of numeric prediction);
• Association rule mining (frequent itemsets and association rules, finding frequent itemsets, generating association rules, filtering association rules);
• Clustering (partitioning methods, hierarchical clustering, density-based clustering, grid-based clustering, model-based clustering, outlier detection);
• Advanced applications;
• Recent trends.

Textbook: MJ. Han and M. Kamber, "Data Mining: Concepts and Techniques," Second Edition, Morgan Kaufmann, 2006. ISBN 1-55860-901-6

Brief description of the type of instruction and learning activities: In accord with the course objectives, the course provides three kinds of learning experiences:

• Class lectures and readings. The lectures cover a broad range of topics on data-mining concepts and techniques. The textbooks provide a more detailed coverage of topics. The next important source of information is the lecture notes.
• Lab and programming exercises. A programming project and assignments explore the ways of implementing data mining concepts and techniques. The project and assignments require the use of tools and skills learned in this course and programming skills in a programming language, e.g. Java, C, or C++.
• Research exercise. Students are asked to research in the literature a data-mining subject and report the findings in a well-structured and well-written paper.

Assignments: Practical exercises include: a programming project, in which students design, implement, test, and evaluate data-mining techniques; 2-3 programming assignments using commercial data mining tools; a term paper on a topic selected in consultation with the instructor; and quizzes, which are given almost every week.

Grading: Grade is assigned cumulatively based on the acquired grades on practical exercises, quizzes, midterm exam, and final exam.