EURO 2025 Leeds
Abstract Submission

1418. Comparing traditional and machine learning methods on company cash flow forecasts in the logistics industry

Invited abstract in session TA-38: Forecasting, prediction and optimization 1, stream Data Science meets Optimization.

Tuesday, 8:30-10:00
Room: Michael Sadler LG19

Authors (first author is the speaker)

1. Alice Wolfe
IESEG School of Management
2. Sarah Van der Auweraer
IESEG School of Management
3. Kristof Coussement
IESEG School of Management

Abstract

Company cash flow management is a critical operation that impacts daily operations, investments, revenue and risk management. With enhanced cash flow forecasting companies can improve their cash management, becoming more agile in their investment decisions and operations, which can bring significant increases in revenue.

Despite its high importance to any operating company, it remains a limited field of study within academic forecasting research and has largely focused on the utilization of traditional forecasting methods. Cash flow forecasting lags behind general financial forecasting in the exploration of Machine Learning methods and their potential improvement to the forecast. The aim of this research is therefore to inspect the application of Machine Learning methods on company cashflow forecasting compared to traditional forecasting methods.

This research is carried out in close collaboration with an international logistics company in the fashion and lifestyle industry. The company provided a data set of approximately 8 million financial records comprised of client and supplier invoice and payment records from 2016 to 2025. This context-specific data is combined with factors regarding contextual client and company data. Using this data, we test and compare traditional forecasting methods: ARIMA and ETS, with Machine Learning methods: LSTM, SVR and Gradient Boosting, to provide a forecast of the company’s cash flow across different time horizons.

Keywords

Status: accepted

Back to the list of papers