ResearchBib Share Your Research, Maximize Your Social Impacts
Sign for Notice Everyday Sign up >> Login

Soil classification and crop cultivation prediction: a comparative study of machine learning models

Journal: International Journal of Advanced Technology and Engineering Exploration (IJATEE) (Vol.11, No. 117)

Publication Date:

Authors : ; ;

Page : 1143-1168

Keywords : Soil classification; Crop cultivation prediction; Support vector machine; Random forest; Multilayer perceptron.;

Source : Downloadexternal Find it from : Google Scholarexternal

Abstract

Soil classification is necessary to optimize agricultural productivity, manage land use effectively, and protect the environment. The soil features, including organic matter, chemical properties, composition, and other organisms, collectively determine its type and enable its differentiation from other soil types. The problem emerges from poor soil type, directly impacting agricultural crop cultivation and presenting a fundamental issue in farming practices. In response to these concerns, this work classified soil based on chemical features such as pH levels, salinity, organic matter, nitrogen, phosphorus, sulphur, and boron, which aim to refine the soil characterization process. There are many lands in Bangladesh where the people think that cultivation is not possible at all. Due to these reasons, many lands are uncultivated in our country every year. Considering these issues, this research aims to predict crop cultivation for a particular soil so that people can select and cultivate various crops undoubtedly and correctly on their land. The soil types are determined first based on their distinct characteristics for a particular zone, then the crop cultivation is ascertained according to the soil types. Various machine learning (ML) techniques such as support vector machine (SVM), decision tree (DT), multilayer perceptron neural network, random forest (RF), logistic regression (LR), and naïve Bayes (NB) are used for prediction. The ML models are selected through a literature-informed process and intended for comparative analysis to determine the most effective techniques. A comparative analysis among different techniques is performed based on performance metrices. The results indicate that the RF algorithm is the most effective for soil classification due to its outstanding accuracy of 96.48%. For crop cultivation prediction, the SVM model outperforms other models with an accuracy of 94.95%. The outcomes of this research endeavour serve as a valuable tool for enhancing farming practices and making substantial contributions to the economy's growth.

Last modified: 2024-09-04 16:16:22