كنفرانس بين المللی مهندسی صنايع و سيستم ها

صفحه اصلی / دوازدهمین كنفرانس بين المللی مهندسی صنايع و سيستم ها

A Data-Driven Framework for Combination Meal Design in Fast-Food Restaurants Using Market Basket Analysis and Association Rule Mining

نویسندگان :

Ali Nokar¹ Amirmohammad Khani² Seyedeh Arefeh Madani-Talab³ Hamidreza Koosha⁴

1- دانشگاه فردوسی مشهد 2- دانشگاه فردوسی مشهد 3- دانشگاه فردوسی مشهد 4- گروه مهندسی صنایع، دانشکده مهندسی، دانشگاه فردوسی، مشهد، ایران

کلمات کلیدی :

Market Basket Analysis،Association Rule Mining،Menu Engineering،Product Attractiveness،Strategic Product Matrix،Restaurant Analytics

چکیده :

Restaurant point-of-sale (POS) data contain rich behavioral information that can support menu optimization, product bundling, and customer-oriented decision-making beyond conventional sales reporting. This study proposes a data-driven framework that integrates market basket analysis, association rule mining, product-level behavioral evaluation, and customer analysis to transform real restaurant transactions into actionable menu and combo decisions. The framework was evaluated using more than 600,000 POS order records, which were parsed into approximately 1.4 million item-level observations and standardized into 87 analyzable menu items. Apriori and FP-Growth were applied under identical configurations and produced exactly the same 1,448 frequent itemsets and 589 association rules, providing an implementation-consistency check for the pattern-mining pipeline. The extracted rules revealed a clear trade-off between statistical strength and commercial reach: high-lift associations were generally concentrated at low support levels, whereas high-support food-beverage combinations provided broader candidates for combo design and point-of-sale prompts. To extend conventional co-purchase analysis, two product-level behavioral metrics, Product Attractiveness Rate and First-Purchase Penetration, were introduced to capture repeat-selection behavior and first-order product prevalence. These metrics were jointly incorporated into a Strategic Product Matrix that classifies menu items into actionable categories for product positioning and menu decision support. An RF-based behavioral segmentation was additionally used to support customer-oriented suggestions and identify customers with reduced purchasing activity. Overall, the proposed framework demonstrates how transaction-level associations and longitudinal purchasing behavior can be integrated into an interpretable decision-support approach for restaurant product intelligence and menu optimization.