Naive Bayesian classifier and random forest approaches for modeling the electrical resistivity of soils in tropical zones by meteorological variables: case of nine sites in Lomé, Togo
1 Department of Electrical Engineering, Polytechnic School of Lomé (EPL), University of Lomé, Togo.
2 Engineering Sciences Research Laboratory (LARSI), University of Lomé, Togo.
3 Department of Electrical Engineering, Higher Technical Training Institute (IFTS) of Lomé, Togo.
Research Article
World Journal of Advanced Research and Reviews, 2023, 20(03), 037–050
Article DOI: 10.30574/wjarr.2023.20.3.2455
Publication history:
Received on 20 October 2023; revised on 28 November 2023; accepted on 01 December 2023
Abstract:
The work combined in this article presents the results of modeling the electrical resistivity of soils based on meteorological data such as geo-referenced coordinates (A), the state of nature the day before (B), the state of nature of the day (C), the ambient temperature (D). A total of 9815 data were sampled over three consecutive years in Lomé, Togo. As methods, we carried out the characterization of the electrical resistivities of the soils measured by Wenner – Schlumberger techniques on nine sites selected in Lomé. Random forests and Naive Bayesian Classifiers are the algorithms used. Certain performance evaluation criteria most commonly encountered in the bibliography are taken into account to evaluate the models. The best result is obtained with random forests and gives MAPE = 17.372%, RMSE = 22.419%, RRMSE = 15.185% and R2 = 70.4%. The result obtained with the naive Bayesian classifier is: MAPE = 24.01%, RMSE = 49.79%, RRMSE = 33.63% and R2 = 37.34%. We deduce from these results that random forests are well suited to predicting the electrical resistivity of soils in tropical areas using meteorological variables. However, it would be good to explore other algorithms to check if the performance will not be better.
Keywords:
Naive Bayesian Classifier; Random Forests; Soil Electrical Resistivity; Classification; Prediction.
Full text article in PDF:
Copyright information:
Copyright © 2023 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0