Answers are grounded in this paper and cite their sources.
Due to the fact that the autocorrelation of time series of passenger demand under normal conditions is, as a rule, practically undeveloped, traditional forecasting methods based on taking into account autocorrelation dependences are not effective enough. The article proposes a direct accounting of the main factor affecting the accuracy of forecasting, namely the factor of seasonal heterogeneity of demand. This accounting is made on the basis of polynomial regression for the time dependence of demand. A specific design example demonstrates the comparative advantages of this approach to assessing the forecast of demand for rail transport.The regression approach is applied to the weekly averaged demand metrics for the time domain, where these metrics are considered known from the sales history. If there is a weekly demand heterogeneity in the forecast zone, an algorithm is proposed to restore such heterogeneity from the initial data.The forecast accuracy based on the proposed method is compared with the results achieved on the basis of the ARIMA model, which reveals, according to preliminary estimates, fairly high accuracy parameters. It is shown on the calculated examples that for the series of demand, which can be considered typical for the sphere of passenger traffic, the regression approach gives the forecast accuracy higher than the ARIMA model. The reasons are considered, due to which, for typical series of passenger demand, the regression approach can be considered as more promising than methods that include taking into account autocorrelation.
No takes yet. Share an insight, caveat, or question.
Венедиктов et al. (2021) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: