MUMBAI, India, May 29 -- Intellectual Property India has published a patent application (202641044400 A) filed by Eswaramuthu M; Kevin Benjamin Samuel; S Sarvesh; and Deepa R, Chennai, Tamil Nadu, on April 7, for 'ai based end to end supply chain: demand forecasting and inventory management.'
Inventor(s) include Eswaramuthu M; Kevin Benjamin Samuel; S Sarvesh; and Deepa R.
The application for the patent was published on May 29, under issue no. 22/2026.
According to the abstract released by the Intellectual Property India: "This invention is related to a forecasting and optimization solution powered by artificial intelligence, which is customized for the retail and supply chain industry. The system combines sophisticated machine learning algorithms and techniques, such as LSTM neural network and XGBoost, to develop highly accurate short-term forecasts for various products sold at multiple store locations. By combining the benefits of LSTM and XGBoost, the invention allows for creating highly accurate forecasts under changing market conditions. The key component of the invention is historical sales data, which is enriched by feature engineering. Besides basic information about product sales, several contextual features, such as product price, promotion status, seasonality effects, store, and product category ID, are used. Time-based variables, like day of the week, month, day, and year, along with rolling demand statistics, such as moving average of 7 and 30 days, provide more accurate forecasts, since the model accounts for seasonal patterns and short-term changes in customer demand. Hybrid forecast mechanism is employed to aggregate forecasts generated by both the LSTM and XGBoost models in order to increase accuracy. The system employs preprocessing and scaling modules to maintain data integrity and improve forecasting performance. The performance of models will be measured based on performance measures such as MAE and RMSE to improve reliability. Apart from generating forecasts for demand predictions, the system is equipped with an inventory optimization tool for translating forecasts into action. Based on comparison between forecast demand and the available inventory, the application gives a recommendation whether the retailer should reorder more items, procure less or maintain current inventory levels. Additionally, more sophisticated inventory control mechanisms can be included such as safety stock computation and reorder point determination. Moreover, a user-friendly interface is provided for the invention by integrating an application built with Streamlit to provide an interactive way of interacting with the system. Users can input data, receive demand forecasts and visualize trends while at the same time monitor their inventory statuses. The system allows for scenario-based forecasting within a specified period (e.g., 30 days)."
Disclaimer: Curated by HT Syndication.
