Demand Forecasting for Domestic Air Transportation in Turkey using Artificial Neural Networks

Nowadays, the competition between companies is rapidly increasing in every industry. This leads to companies trying to be prepared for the near future by forecasting business conditions. The estimated success rate in this context directly affects the success rate of the companies. Airline transport in Turkey, which has grown at a higher rate than Europe's, is an important part of the country's economy and transportation infrastructure. Furthermore, airports encourage development by motivating the commercial activities around them. In the competitive environment of airline transportation, successful forecasting is a crucial issue. Different methods such as multiple linear regression analysis, back-propagation neural networks (BPN), gravity models, multimode models, time series models are used in forecasting studies. In this study, an Artificial Neural Network (ANN) model is used for demand forecasting in domestic air transport in Turkey. In the scope of this study, AzureML, RScript and MATLAB were used for the dataset that is gained between 01.01.2007 - 01.11.2015 and some successful results were obtained. Pearson's correlation coefficient is used as the performance criteria for evaluation and it is observed that the results obtained from the proposed model are at an acceptable level which are gained between 0,79 and 0,93. Therefore, the proposed Artificial Neural Network (ANN) model can be used as a demand forecasting in many areas such as capacity planning, airport infrastructure planning, airplane investments in air transportation.