A Reproducibility Study of Subgroup Discovery Algorithms
摘要
Subgroup discovery is an important data mining technique that aims to uncover specific segments of data that exhibit noteworthy patterns or behaviors. Despite the existence of several subgroup discovery algorithms, there is no thorough comparative analysis of their effectiveness. This paper evaluates two recent subgroup discovery algorithms against the seminal PRIM method, demonstrating PRIM’s superior accuracy, speed, and effectiveness in identifying interesting subgroups. These findings highlight the need for a comprehensive evaluation of subgroup discovery techniques to determine the state of the art.