Algorithms for action rule mining

This work aims at implementing action rules mining in the Python programming languagesince such implementation is not yet available. The thesis also describes a proposal for anexperimental algorithm for generating classification rules from RandomForest models, whichwas designed as an alternative for using class association rule mining for large datasets.The code of the implementation is designed to allow future extensions. The package is sharedwith the data science community and is easy to install. The application has a Scikit-learn likeinterface. The thesis can also serve as a manual for using the created package or as a guide forresearchers who would like to extend it, providing also guide to action rule discovery. The textalso includes performance evaluation with the application LISp-Miner, which already supportsaction rules mining. The implemented solution has performance comparable to LISp-Miner.