Generic Association Rule Engine Optimization Using Item Support Projection

Transactional data analysis and association rule learning have become prominent requirements for effective decision making in marketing operations. Considering that, we propose an optimized near-real-time item associations generation mechanism combined with database utilities. Our solution adopts the association measures of the Apriori algorithm and constructs a live item associations database fed by a generic rule engine. The study also covers redundancy avoidance, memory management and rare event handling with logarithmic transformation. With an experimental case study, we demonstrate hyper-parameter tuning methods and evaluate the results for various scenarios.