IoT Aware Polynomial Regressive Ensemble Artificial Intelligence Model for Crop Yield Prediction in Cloud Computing Environment

Abstract

Accurately estimating crop yields across large geographical regions is essential for ensuring food security and promoting sustainable development. The machine learning driven Artificial Intelligence (AI) based approaches have emerged for enhancing the precision of agricultural yield predictions. While many existing models have achieved higher accuracy by increasing their complexity, their ability to generalize remains limited due to variations in key features across different regions. To address this issue, IoT aware Polynomial Regressive Ensemble AI (IoT-PREAI) Model is introduced. IoT-PREAI technique is to perform several processes. In data harvesting process, the numbers of data samples are collected from the Crop Yield Prediction Dataset. In order to increase the size of input data samples, Adversarial scaling model is employed in IoT-PREAI technique. Data scrubbing is performed to clean the dataset by handling missing data using polynomial function and removing outliers. After that, feature selection is conducted using censored regression method by Roger and Tanimoto Distributed Feature Engineering. Following this, crop yield prediction is performed using the selected features through Adaptive Gradient weight preserved boosting technique. This method processes the data samples and generates the final crop yield prediction results while reducing the error using Nesterov Accelerated Gradient. Experimental evaluation considers several factors. The quantitatively analyzed results indicate that the proposed IoT-PREAI technique achieves higher crop yield prediction accuracy with minimal computation time, RMSE compared to conventional ensemble techniques.