An application on electric vehicle demand forecasti̇ng

In this thesis, data from the period of January 2016 to April 2024 were used to forecast the demand for electric vehicles. During the analysis process, two different methods were applied: Artificial Neural Networks (ANN) and the Autoregressive Integrated Moving Average (ARIMA) model. According to the obtained performance results, the ANN model implemented in R demonstrated a remarkably high level of accuracy, with a coefficient of determination (R²) of 0.959 and a mean absolute error (MAE) of 0.0105. On the other hand, the ARIMA model implemented in the EViews environment yielded an R² value of 0.923 and an MAE value of 0.1966. These results indicate that although both methods possess strong forecasting capabilities, the artificial neural network outperforms the ARIMA model in terms of error rates. Therefore, it is concluded that the ANN method implemented in R is a more suitable option compared to the time series analysis conducted in EViews. This study contributes significantly to the literature by examining both time series analysis and artificial neural network methods in demand forecasting. The findings provide a methodological framework that can be utilized in solving similar problems and offer guidance for future research in this field. Additionally, by comparing traditional and AI-based models, the study aims to serve as a reference for researchers in critical areas such as model selection and forecasting accuracy. In this respect, the study holds the potential to serve as a foundational reference for the development of new methods and applications in future research