MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611106904 A) filed by Panipat Institute Of Engineering And Technology Piet on September 04, 2026, for System And Method For Explainable Multi-Variate Agricultural Yield Prediction.
Inventors include Dr. Gurmeet Kaur; Dr. Sakshi Patni; and Prof. Dr. Dinesh C. Verma.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: [048] The present invention provides a sensor-coupled system and method for explainable multi-variate agricultural yield prediction. A field transducer array acquires soil pH, macronutrient, soil-quality and micro-climatic parameters and communicates the parameters to a processing node. The node performs structural validation, bounded normalization and staged feature selection using significance evaluation, correlation analysis and impurity-based ranking. A selected feature set is processed by a gradient-boosting regression ensemble and a recurrent neural-network processor. Predictive outputs are verified using multiple cross-validation strata, error criteria and residual diagnostics. An explainability engine determines feature-level contributions and generates an interpretable deployment vector comprising a validated crop- yield prediction and associated attribution information. The deployment vector may be supplied to precision-agriculture decision support, variable-rate agricultural machinery, cloud agronomy services or agricultural risk systems, thereby providing a traceable machine-actionable output from sensed field conditions. Accompanied Drawing [FIGS. 1-2]
Disclaimer: Curated by HT Syndication.
