MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202621081985 A) filed by Symbiosis International Deemed University on July 02, 2026, for Ensemble Machine Learning Based Web System For Real-Time Stock Index Forecasting Using Hybrid Arima-Lstm-Gbrt Models.
Inventors include Dr. Deepak Suresh Asudani; Shubham Chakraborty; Siddhant Bahekar; and Sumedh Yete.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: ABSTRACT ENSEMBLE MACHINE LEARNING BASED WEB SYSTEM FOR REAL-TIME STOCK INDEX FORECASTING USING HYBRID ARIMA-LSTM-GBRT MODELS The present invention discloses an ensemble machine learning based web system (100) for real-time stock index forecasting. The system (100) comprises a data acquisition and preprocessing module (110) for retrieving and processing historical stock market data with technical indicator feature engineering, a back-end computational engine (120) integrating an ARIMA model component (130) for linear trend capture, an LSTM model component (140) with gating mechanisms for non-linear temporal dependency modelling, and a GBRT model component (145) for robust regression through sequential decision tree construction. An ensemble aggregation engine (150) combines predictions from the three model components through dynamically weighted averaging based on backtesting performance. A performance evaluation module (155) computes accuracy metrics including MSE, RMSE, MAE, MAPE, and R-squared. An interactive web-based user interface module (160) enables users to input ticker symbols, select models, configure forecast horizons, and visualize results through interactive charts and performance tables. [
Disclaimer: Curated by HT Syndication.
