Measures of interestingness of association rules

The master thesis focuses on the topic of interestingness measures of association rules. The main goal of this thesis is to collect some widely used measures of interestingness of association rules from the literature and create a Python package that implements the most important ones to be used in Cleverminer. The thesis also includes three additional goals. The first additional goal of this thesis is to show how the user can add their own measures. The second additional goal is to apply selected measures on a publicly available dataset and compare the performance of the measures with the theoretical claims made about the measures in the literature. The last goal is to measure the performance using these custom metrics in Cleverminer. A comprehensive review of widely used interestingness measures is conducted and the most important measures are implemented in a python package called miar (available at: https://github.com/M-Benes/miar or https://pypi.org/project/miar/) which is published on PyPI and globally available to install, with complete source code available on GitHub for users to use, as well as prepared to be included in Cleverminer. The performance of the package is tested, and users can also modify the package and even add their own custom measures if the package is installed locally. A selection of measures is applied to a real dataset, and their performance is compared to theoretical claims about the measures in the literature. The results of the thesis include the miar package, a complex table with over 100 interestingness measures and over 35 sources, a performance test to prove the effectiveness of the package, and a discussion about the performance of a selection of the measures.