Characteristics of the application of arima-SVM methods in the forecasting of non-scheduled passenger air transportation

The article examines the independent application of ARIMA (Auto Regressive Integrated Moving Average) and SVM (Support Vector Machine) methods for forecasting non-scheduled passenger air transportation. Based on the SVM model, the data is classified based on different kernel functions, and the best prediction results are determined. The autoregression (ARIMA) model is applied to identify linear trends and regularities within time series data. Based on the analysis, the results show that ARIMA and SVM models offer superior forecasting accuracy and reliability. Nevertheless, the relative error of the prediction results compared to the actual indicators is smaller in the SVM model. This also shows that the identification of non-linear relationships between data in non-scheduled passenger air transportation makes forecasting results more effective and optimal. The obtained results will serve as an effective tool in forecasting the demand for non-scheduled passenger air transportation. As a result, substantial support will be observed in the planning of operations in the mentioned field, preparation of the existing infrastructure according to the demand, etc.