Location Specific Paddy Yield Prediction using Monte Carlo Simulation incorporated Long Short-Term Memory

Abstract

Accurately predicting paddy yield is vital for food security and efficient farm management. This work proposes LSPYP-ML, a framework that combines fuzzy logic, Monte Carlo simulation, and Long Short-Term Memory (LSTM) networks to improve prediction accuracy. The fuzzy module cleans and classifies uncertain data such as rainfall, temperature, and pesticide use. The Monte Carlo module simulates extreme weather scenarios to account for environmental variability. Finally, the LSTM module captures temporal patterns in climate and yield data for robust forecasting. Experiments show that the framework achieves higher accuracy, precision, sensitivity, specificity, and F-Score compared to existing methods. LSPYP-ML offers a reliable decision-support tool for farmers and policymakers to enhance productivity and manage climate risks.