Mint: MDL-based approach for Mining INTeresting Numerical Pattern Sets

Abstract Pattern mining is well established in data research, especially for binary datasets. Surprisingly, there much less work about numerical pattern and this research area remains under-explored. In paper we propose Mint , an efficient MDL-based algorithm The MDL principle a robust reliable framework widely used mining, as subgroup discovery. reuse discovering useful patterns returning set of non-redundant overlapping with well-defined boundaries covering meaningful groups objects. not alone the category miners based on MDL. experiments presented show that outperforms competitors among which IPD, RealKrimp Slim .