An adaptive neural network-fuzzy linear regression approach for improved car ownership estimation and forecasting in complex and uncertain environments: the case of Iran

Abstract

This paper applies a novel adaptive approach consisting of Artificial Neural Network (ANN) and Fuzzy Linear Regression (FLR) to improve car ownership forecasting in complex, ambiguous, and uncertain environments. This integrated approach is applied to forecast car ownership in Iran from 1930 to 2007. In this study, the level of car ownership is viewed as the result of demographic, politico-social, and urban structure factors including average family size, total population density, urban population density, urbanization rate, gross national product per capita, gasoline price, and total road length. To capture the potential complexity, uncertainty, and linearity relation between the car ownership function and its determinants, ANN and FLR (including eight well-known FLR) approaches are applied to the collected data. Next, the preferred ANN is selected based on sensitivity analysis results for the test data while the preferred FLR is identified with regard to ANOVA and MAPE results. The results obtained from the performance comparison demonstrate the considerable superiority of the preferred ANN over the preferred FLR regarding the nonlinear and complex nature of the car ownership function in Iran. This is the first study that presents an ANN-FLR approach for car ownership forecasting capable of handling complexity and non-linearity, uncertainty, pre-processing, and post-processing.

Keywords:

• forecasting
• car ownership
• optimization
• artificial neural network (ANN)
• fuzzy linear regression (FLR)
• complexity
• uncertainty
• sensitivity analysis
• Iran

Acknowledgements

The authors are grateful for the valuable comments and suggestion from the respected reviewers. Their valuable comments and suggestions have enhanced the strength and significance of our paper.

The authors would like to acknowledge the financial support of University of Tehran for this research under grant number 8106013/1/04.