Highlights Combining time series with machine learning models reduces prediction error. Predictions using RL committee of forecasters closely track market changes. Long term weekly forecasts from monthly values outperform direct weekly forecasts.
Abstract Freight forecasting is essential for managing, planning operating and optimizing the use of resources. Multiple market factors contribute to the highly variable nature of freight flows, which calls for adaptive and responsive forecasting models. This paper presents a demand forecasting methodology that supports freight operation planning over short to long term horizons. The method combines time series models and machine learning algorithms in a Reinforcement Learning framework applied over a rolling horizon. The objective is to develop an efficient method that reduces the prediction error by taking full advantage of the traditional time series models and machine learning models. In a case study applied to container shipment data for a US intermodal company, the approach succeeded in reducing the forecast error margin. It also allowed predictions to closely follow recent trends and fluctuations in the market while minimizing the need for user intervention. The results indicate that the proposed approach is an effective method to predict freight demand. In addition to clustering and Reinforcement Learning, a method for converting monthly forecasts to long-term weekly forecasts was developed and tested. The results suggest that these monthly-to-weekly long-term forecasts outperform the direct long term forecasts generated through typical time series approaches.
Reinforcement learning framework for freight demand forecasting to support operational planning decisions
2020-03-21
Article (Journal)
Electronic Resource
English
British Library Conference Proceedings | 2009
|