Optimizing Fresh Food Supply Chains Through Correlation Insights and LSTM Forecasting

Managing the short shelf life and variable freshness of vegetables in large fresh food supermarkets poses significant challenges, often leading to discounted sales or disposal of unsold products, thereby reducing profits. To address these issues, supermarkets leverage data-driven approaches to optimize replenishment and pricing strategies. This paper investigates the complexities of vegetable sales and supply chain management, proposing a comprehensive predictive and decision-making model utilizing deep learning techniques. Our contributions are threefold: Firstly, we apply the Fourier transform to analyze sales data and identify seasonal patterns in vegetable sales. Secondly, we use Kolmogorov-Smirnov to check the distribution of data to judge the single sale of a kind of vegetables. Meanwhile, we employ Spearman rank correlation analysis to assess the interdependence of different vegetable sales, enabling the formulation of improved bundling and promotional strategies. Thirdly, we develop an LSTM forecasting model to predict future sales based on historical data, facilitating more effective restocking decisions. At last, we introduce the Time Series Cross-Validation method to evaluate the generalization ability and stability of the model for predicting future vegetable sales. These methodologies aim to enhance profitability and customer satisfaction by equipping supermarkets with advanced tools for managing fresh produce.