A Multi-Level Association Rule Mining Algorithm Based on NSGA-II for Market Basket Analysis

As online ordering gradually replaces the traditional offline dining model, more and more businesses are using online systems to sell their products, leaving behind a large amount of order information. By mining the correlations between products from a large amount of order information and discov-ering customers’ purchasing behavior patterns, businesses can develop effective marketing strategy, enhance customer loyalty and improve economic benefits. Traditional precise algorithms such as Apriori and FP-growth algorithms are widely used for multi-level association rule mining, but they have problems with high computational complexity and space occupation. To solve these problems, intelligent evolution algorithms have been introduced. Considering the limitations of single-objective evo-lutionary algorithms that they can only optimize one objective, this paper simultaneously considers multiple optimization indi-cators, transforms the Market Basket Analysis problem into a multi-objective optimization problem, and proposes a multi-level association rule mining algorithm based on NSGA-II. At each level, we first use the Apriori algorithm to obtain frequent itemset which are then utilized for population initialization in NSGA-II and then use the NSGA-II algorithm for shopping basket data association rule mining. In addition, we have also designed an acceleration module to improve the computational efficiency of the algorithm. We applied this algorithm to actual production data from a coffee shop and the experimental results show that the algorithm is efficient in processing large volumes of orders and can effectively mine high-quality rules.