Forecasting container freight rates using the Prophet forecasting method

Abstract

This study applies three innovative methods in forecasting container freight rates. Firstly, we extracted 471 major disruptive events from the ‘Lloyds List’ database from 2010 until 2020, that may affect freight rates. Secondly, we use Machine Learning (ML) and natural language processing techniques to categorize these events into six distinct categories. These include: “congestion”, “peak demand”, “policy”, “price up”, “overcapacity”, and “coronavirus”. Thirdly, we apply Prophet forecasting on six major container routes by incorporating the six categories of events. The results reveal that ‘overcapacity’ and ‘coronavirus’ led to improved forecasting accuracy of freight rates when compared to the Prophet model without accounting for events. This study provides a more reliable mixed-method approach to improving the accuracy of container freight rate forecasts. The proposed forecasting technique will help policy makers and practitioners to develop and deploy strategies to mitigate the risks associated with the volatility of freight rates and supply chain costings.