Discovering outstanding subgroup lists for numeric targets using MDL

The task of subgroup discovery (SD) is to find interpretable descriptions of subsets of a dataset that stand out with respect to a target attribute. To address the problem of mining large numbers of redundant subgroups, subgroup set discovery (SSD) has been proposed. State-of-the-art SSD methods have their limitations though, as they typically heavily rely on heuristics and/or user-chosen hyperparameters. We propose a dispersion-aware problem formulation for subgroup set discovery that is based on the minimum description length (MDL) principle and subgroup lists. We argue that the best subgroup list is the one that best summarizes the data given the overall distribution of the target. We restrict our focus to a single numeric target variable and show that our formalization coincides with an existing quality measure when finding a single subgroup, but that-in addition-it allows to trade off subgroup quality with the complexity of the subgroup. We next propose SSD++, a heuristic algorithm for which we empirically demonstrate that it returns outstanding subgroup lists: non-redundant sets of compact subgroups that stand out by having strongly deviating means and small spread.

[1]  Arno J. Knobbe,et al.  Diverse subgroup set discovery , 2012, Data Mining and Knowledge Discovery.

[2]  Willi Klösgen,et al.  Explora: A Multipattern and Multistrategy Discovery Assistant , 1996, Advances in Knowledge Discovery and Data Mining.

[3]  Jilles Vreeken,et al.  Identifying consistent statements about numerical data with dispersion-corrected subgroup discovery , 2017, Data Mining and Knowledge Discovery.

[4]  Jilles Vreeken,et al.  Krimp: mining itemsets that compress , 2011, Data Mining and Knowledge Discovery.

[5]  Chedy Raïssi,et al.  Anytime discovery of a diverse set of patterns with Monte Carlo tree search. (Découverte d'un ensemble diversifié de motifs avec la recherche arborescente de Monte Carlo) , 2017 .

[6]  Thomas Bäck,et al.  Identifying flight delay patterns using diverse subgroup discovery , 2018, 2018 IEEE Symposium Series on Computational Intelligence (SSCI).

[7]  Tijl De Bie,et al.  Subjectively Interesting Subgroup Discovery on Real-Valued Targets , 2017, 2018 IEEE 34th International Conference on Data Engineering (ICDE).

[8]  Teemu Roos,et al.  Minimum Description Length Revisited , 2019, ArXiv.

[9]  Luis Nunes,et al.  Hotel booking demand datasets , 2019, Data in brief.

[10]  Mehdi Kaytoue-Uberall,et al.  FSSD - A Fast and Efficient Algorithm for Subgroup Set Discovery , 2019, 2019 IEEE International Conference on Data Science and Advanced Analytics (DSAA).

[11]  Peter A. Flach,et al.  Subgroup Discovery with CN2-SD , 2004, J. Mach. Learn. Res..

[12]  Ronald L. Rivest,et al.  Learning decision lists , 2004, Machine Learning.

[13]  J. Rissanen,et al.  Modeling By Shortest Data Description* , 1978, Autom..

[14]  Martin Atzmüller,et al.  Subgroup discovery , 2005, Künstliche Intell..

[15]  Matthijs van Leeuwen,et al.  Interpretable multiclass classification by MDL-based rule lists , 2019, Inf. Sci..

[16]  Marvin Meeng,et al.  For real: a thorough look at numeric attributes in subgroup discovery , 2020, Data Min. Knowl. Discov..

[17]  Matthijs van Leeuwen,et al.  Maximal exceptions with minimal descriptions , 2010, Data Mining and Knowledge Discovery.

[18]  Jorma Rissanen,et al.  Minimum Description Length Principle , 2010, Encyclopedia of Machine Learning.

[19]  Wynne Hsu,et al.  Integrating Classification and Association Rule Mining , 1998, KDD.

[20]  Esther Galbrun,et al.  Association Discovery in Two-View Data , 2015, IEEE Transactions on Knowledge and Data Engineering.

[21]  Margo I. Seltzer,et al.  Scalable Bayesian Rule Lists , 2016, ICML.

[22]  Stefan Rüping,et al.  On subgroup discovery in numerical domains , 2009, Data Mining and Knowledge Discovery.

[23]  W. Johnson,et al.  The Bayesian Two-Sample t Test , 2005 .

[24]  Jeffrey N. Rouder,et al.  Bayesian t tests for accepting and rejecting the null hypothesis , 2009, Psychonomic bulletin & review.