Reducing False Discoveries in Statistically-Significant Regional-Colocation Mining: A Summary of Results

TL;DR

This work addresses the problem of false discoveries in statistically significant regional colocation pattern mining by introducing MultComp-RCM, a Bonferroni-corrected miner that builds on the prior SSRCM approach. MultComp-RCM reduces the experiment-wide false positive rate while lowering computational cost through adjusted per-test thresholds and sequential union of atomic partitions. The authors provide theoretical guarantees comparing Type-I error and cost to SSRCM, and validate the method with synthetic data and a Minnesota SafeGraph case study, illustrating new regional patterns among retail brands and supporting reductions in false discoveries. The results demonstrate practical impact for spatial data mining by delivering more reliable regional patterns with improved efficiency, and point to future work on balancing error types and adding temporal dimensions. The work contributes a principled, scalable framework for significance-aware regional colocation mining with broad applicability to ecology, economics, and sociology.

Abstract

Given a set \emph{S} of spatial feature types, its feature instances, a study area, and a neighbor relationship, the goal is to find pairs $<$a region ($r_{g}$), a subset \emph{C} of \emph{S}$>$ such that \emph{C} is a statistically significant regional-colocation pattern in $r_{g}$. This problem is important for applications in various domains including ecology, economics, and sociology. The problem is computationally challenging due to the exponential number of regional colocation patterns and candidate regions. Previously, we proposed a miner \cite{10.1145/3557989.3566158} that finds statistically significant regional colocation patterns. However, the numerous simultaneous statistical inferences raise the risk of false discoveries (also known as the multiple comparisons problem) and carry a high computational cost. We propose a novel algorithm, namely, multiple comparisons regional colocation miner (MultComp-RCM) which uses a Bonferroni correction. Theoretical analysis, experimental evaluation, and case study results show that the proposed method reduces both the false discovery rate and computational cost.

Figures (10)

• Figure 1: Regions where all or subsets of $f_A$, $f_B$ and $f_C$ significantly co-locate in the study area
• Figure 2: Comparison with Related Work
• Figure 3: Two example null hypotheses for significance testing of the observed data in Figure \ref{['fig:ColocationToyExample']}
• Figure 4: Execution trace of SSRCM and MultComp-RCM.
• Figure 5: Overall validation framework

Theorems & Definitions (2)

• Lemma 1
• Lemma 2