Optimization of Vegetable Restocking and Pricing Strategies for Innovating Supermarket Operations Utilizing a Combination of ARIMA, LSTM, and FP-Growth Algorithms

Abstract

In the dynamic environment of fresh food supermarkets, managing the short shelf life and varying quality of vegetable products presents significant challenges. This study focuses on optimizing restocking and pricing strategies to maximize profits while accommodating the diverse and time-sensitive nature of vegetable sales. We analyze historical sales, pricing data, and loss rates of six vegetable categories in Supermarket A from 1 July 2020 to 30 June 2023. Using advanced data analysis techniques like K-means++ clustering, non-normal distribution assessments, Spearman correlation coefficients, and heat maps, we uncover significant correlations between vegetable categories and their sales patterns. The research further explores the implications of cost-plus pricing, revealing a notable relationship between pricing strategies and sales volumes. By employing Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models, we forecast sales and determine optimal restocking volumes. Additionally, we use price elasticity theories and a comprehensive model to predict net profit changes, aiming to enhance profit margins by 47%. The study also addresses space constraints in supermarkets by proposing an effective assortment of salable items and individual product restocking plans, based on FP-Growth algorithm analysis and market demand. Our findings offer insightful strategies for sustainable and economic growth in the supermarket industry, demonstrating the impact of data-driven decision-making on operational efficiency and profitability.

Keywords

• optimizing restocking and pricing strategies
• K-means++ clustering
• ARIMA model
• LSTM model
• FP-Growth algorithm