Forecasting railway freight volume using non-seasonal non-stationary time series models

This study analyzes the time series data of railway freight volume from 1949 to 2008. The original data exhibits an overall upward trend but with significant fluctuations, particularly in the 1960s and early 1970s. Through first-order differencing, the stationarity of the data was improved, and an ARIMA (1,1,0) model was established. The model fits the data adequately, with a significant first-order autoregressive term. Ten-year forecasts indicate that railway freight volume will continue to grow, though uncertainties remain, especially during historically volatile periods. This research demonstrates the applicability and potential of ARIMA models in time series analysis and forecasting, providing support for railway freight planning and decision-making.