Using Artificial Neural Network Models to Analyse Diesel Prices in Selected Regions in Tanzania

Authors

  • Paul Theophily Nsulangi Dar es Salaam Maritime Institute (DMI), Department of Marine Engineering, P.O Box 6727, Dar es Salaam, Tanzania.
  • Miraji Abdallah Mkwande Dar es Salaam Maritime Institute (DMI), Department of Marine Engineering, P.O Box 6727, Dar es Salaam, Tanzania.
  • John Mbogo Kafuku University of Dar es Salaam (UDSM), Department of Mechanical and Industrial Engineering (MIE), College of Engineering and Technology (CoET), P.O. Box P.O Box 35091, Dar es Salaam, Tanzania.
  • Werneld Egno Ngongi Dar es Salaam Maritime Institute (DMI), Department of Marine Engineering, P.O Box 6727, Dar es Salaam, Tanzania.

Keywords:

Artificial Neural Network, Monthly Cap Price, Feed-forward Backpropagation, Cap Prices Prediction.

Abstract

In the current study, artificial neural network (ANN) models are applied to estimate monthly diesel cap prices for the three selected Regions; Mbeya, Ruvuma and Katavi, utilizing data obtained in Mbeya City, Songea and Mpanda Municipalities respectively. The study proposed 5-12-10-1, 5-10- 10-1 and 5-12-8-1 architectures for the Mbeya City ANN model, Songea Municipal ANN model and Mpanda Municipal ANN model, respectively, due to their exceptional estimation capabilities. The performance forecast of the ANN models was assessed with that of the historical monthly diesel cap price published by the EWURA. The results demonstrated that the suggested ANN models achieved R2 and MAE values of 1.0000, 1.0000, 1.000 and 1.31 x 10-12, 1.08 x 10-12, 1.23 x 10-12 for ANN models for Mbeya City, Songea and Mpanda Municipalities, respectively, historical monthly diesel cap prices. Additionally, the study analysed the trends of the monthly diesel cap price variations, utilising outputs of the ANN models. Based on the analysis it shows that from July 2015 to February 2016, the monthly diesel price decreased by an average of 3.41%. Whereas, starting from March 2016 to December 2018, the monthly diesel price increased by an average of 1.56%. The analysis results demonstrate that the suggested ANN models exhibited superior performance in predicting monthly diesel cap prices in the study areas. Therefore, it can be deduced that the proposed ANN model is a reliable and effective tool for analysing monthly diesel cap prices in the selected regions. Based on the results, it can be concluded that the proposed ANN models are accurate and useful tools for analysing monthly diesel prices in Mbeya, Ruvuma and Katavi.

Downloads

Published

04-05-2025

How to Cite

Theophily Nsulangi, P., Abdallah Mkwande, M., Mbogo Kafuku, J., & Egno Ngongi, W. (2025). Using Artificial Neural Network Models to Analyse Diesel Prices in Selected Regions in Tanzania. The Journal of Maritime Science and Technology (JMST), 1(1). Retrieved from https://journal.dmi.ac.tz/index.php/1DMI1/article/view/69

Issue

Section

Articles