The application of Neural Prophet Time Series in predicting rice stock at Rice Stores

Abstract

Efficient inventory management and consistent rice supply are pivotal for the sustainability of small-scale food stalls. This research introduces an innovative approach to address this challenge through the Neural Prophet algorithm. By synergizing neural networks with additive regression models, the Neural Prophet captures intricate temporal patterns and trends within rice sales data. Our study evaluates the Neural Prophet's effectiveness in predicting rice sales, specifically for essential food vendors. Leveraging historical sales data from June 2022 to April 2023, the algorithm incorporates seasonality and trends and integrates external events, such as holidays, to heighten prediction precision. Our findings underscore the Neural Prophet's remarkable prowess in forecasting rice sales at primary food kiosks, adeptly discerning data trends and fluctuations, culminating in reliable future sales projections. The model boasts compelling performance metrics: MAE = 12.90, RMSE = 15.80, and Loss = 0.0313. Beyond its technical merits, this research carries significant practical implications, empowering proprietors and suppliers of basic food stalls to streamline inventory management, avert stockouts, and curtail overstocking by harnessing the precision of rice demand forecasting facilitated by the Neural Prophet algorithm.