Selected data mining methods and their applicability to the television audience monitoring data in the Czech Republic

Abstract

Data mining is nowadays a fast-growing field, which incorporates machine learning, statistics, and logic within computer science. It has the potential to bring new insights into almost all branches of human activity, because the data are stored almost everywhere. This thesis tries to show the main aspects of the original Czech method Guha, to demonstrate its strength via its application to television audience data, and finally to compare it with the association rules method, which is similar to it. The ambition of this text is to interconnect the world of praxis with the theoretical field, where methods are invented. It also serves as an introduction to data mining itself. The results show that Guha is a full-value method with several interesting features and might be a good tool for extracting knowledge from analyzed data.