Title: Fertility level prediction in precision agriculture based on an ensemble classifier model
Authors: Bhuvaneswari Swaminathan; Saravanan Palani; Subramaniyaswamy Vairavasundaram
Addresses: School of Computing, SASTRA Deemed University, Thanjavur, India ' School of Computing, SASTRA Deemed University, Thanjavur, India ' School of Computing, SASTRA Deemed University, Thanjavur, India
Abstract: Nowadays, machine learning in precision agriculture have become a promising approach for increasing productivity without environmental impact. Despite significant technology advancements, lack of applicability in soil health management leads to lower crop yield. Soil testing is a predominant process to determine the quantity of soil nutrients for crop growth. Thus, the scope of this study investigates the land fertility level based on land chemical properties. An ensemble classifier is incorporated to handle multi-class soil nutrients data. The model's overall objective is to reduce wasteful expenditure on the amount of fertiliser to the land, reduce the intervention of soil science experts, and preserve soil health. The performance of the classification model regularises using grid search tuning mechanism. The experiments were conducted on different data set proportions with various metrics. AUC shows that proposed determine high accuracy of 99.33%. It is clear that proposed approach outperforms over existing state-of-art-techniques in fertility level identification.
Keywords: soil nutrients information; soil fertility level; correlation coefficient; random forest classification; hyper parameter optimisation; grid search mechanism.
International Journal of Sustainable Agricultural Management and Informatics, 2021 Vol.7 No.4, pp.270 - 288
Received: 13 Apr 2021
Accepted: 17 Jul 2021
Published online: 07 Apr 2022 *