Mining for Insights: An Evaluation of Association Rules for Household Water Perception Data

Abstract

Understanding household perceptions of water quality and safety is essential for effective water management and public health planning. This study analyses household water perception data from Sri Lanka’s Northern Province using Association Rule Mining (ARM). Data from 101 households covering twelve categorical perception variables were examined using the Apriori algorithm. Frequent itemsets and association rules were evaluated using support, confidence, and lift measures. The analysis identified consistent perception patterns, particularly linking perceived water quality and supply reliability with household precautionary practices. Several strong, high-confidence rules indicated non-random relationships among perception variables. Overall, the findings demonstrate that ARM is a transparent and effective method for uncovering latent perception structures in household survey data and provides useful insights to inform water management and public health interventions.