This study addresses the challenge of dynamic pricing and replenishment planning for fresh vegetables in retail environments, where product shelf life is short and quality dergades rapidly. Using historical sales, wholesale pricing, and spoilage data from July 2020 to June 2023, a multi-phase analytical framework was developed.

The first phase involves analyzing interdependencies among vegetable categories and individual SKUs through correlation analysis. Sales volume distributions across different time granularities—quarterly, monthly, weekly, and hourly—are examined using pivot tables and heatmaps. Time-series decomposition reveals recurring patterns such as weekday peaks and seasonal fluctuations in demand.

For replenishment optimization at the category level, a cost-plus pricing model is calibrated against historical sales elasticity. Demand forecasting models—including neural networks, Gaussian process regression, and decision tree regressors—are trained on price-volume pairs to estimate revenue-maximizing price points. The models are validated using RMSE metrics and visualized via scatter plots comparing actual vs. predicted volumes.

At the SKU level, constraints are applied: total available SKUs must be between 27 and 33, with each item ordered no less than 2.5 kg. A mixed-integer linear programming approach is used to select optimal SKUs and determine quantities that maximize profit while meeting minimum display requirements and spatial limits.

Data enhancements are recommended, including real-time temperature monitoring during transport, customer demographic segmentation, and weather impact modeling. These additions would improve forecast accuracy and support adaptive pricing strategies under supply chain uncertainty.