Integrating statistical significance and discriminative power in pattern discovery

Pattern discovery plays a central role in knowledge acquisition across multiple domains. Actionable patterns must meet rigorous statistical significance criteria and, in the presence of target variables, further uphold discriminative power. This work addresses the underexplored area of guiding pattern discovery by integrating statistical significance and discriminative power criteria into state-of-the-art algorithms while preserving pattern quality. We also address how pattern quality thresholds, imposed by some algorithms, can be rectified to accommodate these additional criteria. To assess the proposed methodology, we select the triclustering task as the guiding pattern discovery case and extend well-known greedy and multi-objective optimization triclustering algorithms, δ -Trimax and TriGen, with various merit functions, such as Mean Squared Residual (MSR), Least Squared Lines (LSL), and Multi Slope Measure (MSL). Results from three case studies show the role of the proposed methodology in discovering patterns with pronouncedly higher discriminative power and statistical significance without quality deterioration, highlighting the relevance of combining these criteria to guide the underlying searches and aid knowledge discovery. Although the proposed methodology is motivated over multivariate time series data, it is straightforwardly extensible to pattern discovery tasks involving multivariate, N-way (N > 3), transactional, and sequential data structures. Availability: The code is freely available at https://github.com/JupitersMight/MOF_Triclustering • Integration of discriminative power views into pattern discovery. • Adaptable framework for pattern-centric knowledge acquisition. • Patterns with guarantees of quality, statistical significance, and discriminative power. • Extended triclustering views for pattern discovery in tensor data. • Mitigation of false positive and false negative risks during pattern discovery.