Forecasting Automobile Sales in Turkey with Artificial Neural Networks

Forecasting Automobile Sales in Turkey with Artificial Neural Networks

Aycan Kaya, Gizem Kaya, Ferhan Çebi
Copyright: © 2019 |Pages: 11
DOI: 10.4018/IJBAN.2019100104
(Individual Articles)
No Current Special Offers


This study aims to reveal significant factors which affect automobile sales and estimate the automobile sales in Turkey by using Artificial Neural Network (ANN), ARIMA, and time series decomposition techniques. The forecasting model includes automobile sales, automobile price, Euro and Dollar exchange rate, employment rate, consumer confidence index, oil prices and industrial production confidence index, the probability of buying an automobile, female employment rate, general economic situation, the expectation of general economic situation, financial status of households, expectation of financial status of households. According to the regression results, changes in Dollar exchange rate, the expectation of financial status of households, seasonally adjusted industrial production index, logarithmic form of automobile sales before-one-month which have a significant effect on automobile sales, are found to be the significant variables. The results show that ANN has a better estimation performance with MAPE=1.18% and RMSE=782 values than ARIMA and time series decomposition techniques.
Article Preview

Literature Review

Demand forecasting methods are divided into two categories as qualitative and quantitative methods. The qualitative methods like Delphi technique, market research, and expert opinions are subjective methods based on one's thoughts and experience, and they are used in uncertain situations where there are insufficient data. On the other hand, the quantitative methods are used in cases where there are sufficient numerical quantities based on mathematical models (Karaatli et al., 2012). Quantitative methods are divided into two categories as time series (Box-Jenkins method, trend analysis, moving averages, exponential smoothing technique, etc.) and mixed methods (simple and multi-regression, econometric models and artificial intelligence and heuristics (genetic algorithm, support vector machines and artificial neural networks)) (Karaatli et al., 2012).

In the literature, several studies are using different demand forecasting methods to handle demand forecasting problem in the automotive sector. Wang et al. (2011) use an adaptive-network-based fuzzy inference system (ANFIS) to estimate new automobile sales in Taiwan. They use coincident indicators, leading indicators, wholesale price indices, independent indices and exchange rates in their forecasting model. The determinants of these indicators are as follows:

  • Coincident indicators: Industrial production index, real customs-cleared exports, the sales of manufacturing, the sales index of wholesale, retail, and food services, real machinery and electrical equipment imports, the employment of non-agricultural, the total power consumption.

  • Leading indicators: Average monthly overtime in industry and services, the index of export orders, the superficial measurements of housing starts and building permits, the indexes of producer's inventory, real monetary aggregates, SEMI book-to-bill ratio, stock price index.

  • Wholesale price indices: The prices of the automobile, the oil prices, the prices of automobile components.

  • Independent indices: Population, unemployment rate, the average earnings of employees in industry and services.

  • Exchange rates: Exchange rates the N.T. dollar against the US dollar, exchange rates for the N.T. dollar against the Eurodollar.

Complete Article List

Search this Journal:
Volume 10: 1 Issue (2023)
Volume 9: 6 Issues (2022): 4 Released, 2 Forthcoming
Volume 8: 4 Issues (2021)
Volume 7: 4 Issues (2020)
Volume 6: 4 Issues (2019)
Volume 5: 4 Issues (2018)
Volume 4: 4 Issues (2017)
Volume 3: 4 Issues (2016)
Volume 2: 4 Issues (2015)
Volume 1: 4 Issues (2014)
View Complete Journal Contents Listing