Crop yield prediction using effective deep learning and dimensionality reduction approaches for Indian regional crops

Abstract

Crop yield prediction (CYP) at the field level is crucial in quantitative and economic assessment for creating agricultural commodities plans for import-export strategies and enhancing farmer incomes. Crop breeding has always required a significant amount of time and money. CYP is developed to forecast higher crop production. This paper proposes an efficient deep learning (DL) and dimensionality reduction (DR) approaches for CYP for Indian regional crops. This paper comprised ‘3′ phases: preprocessing, DR, and classification. Initially, the agricultural data of the south Indian region are collected from the dataset. Then preprocessing is applied to the collected dataset by performing data cleaning and normalization. After that, the DR is performed using squared exponential kernel-based principal component analysis (SEKPCA). Finally, CYP is based on a weight-tuned deep convolutional neural network (WTDCNN), which predicts the high crop yield profit. The simulation outcomes shows that the proposed method attains superior performance for CYP compared to exiting schemes with an improved accuracy of 98.96 %. The novelty of the proposed approach lies in the combination of DL, DR, and WTDCNN techniques for accurate crop yield prediction, specifically tailored for Indian regional crops.

Keywords

• Crop yield prediction
• Machine learning
• Deep learning
• Principal component analysis
• Deep convolutional neural network