Machine Learning Implementation for Demand Forecasting in Supply Chain Management

Yumo Bai

2024

Abstract

This paper aims to enhance demand forecasting accuracy in retail supply chains with the application of machine learning techniques: Autoregressive Integrated Moving Average (ARIMA) and XGBoost models. In this respect, the research addressed shortcomings of traditional approaches to forecasting, very often missing the complexity of modern demand patterns due to their reliance on historical data and simplistic assumptions. While the ARIMA model would model seasonality in time series data, the XGBoost would model more complex and nonlinear interactions among multiple features. The dataset was sourced from Kaggle. Treatment of missing values and outliers was handled, and further enhancement by feature engineering was added. The results indicate that while ARIMA is very effective in capturing temporal dependencies and seasonal trends, XGBoost outperforms it in handling complex relationships that deal with fuel prices and CPI. Indeed, the combination of both models makes for a holistic model toward demand forecasting, illustrating substantial improvements in the accuracy of the forecast. This can be taken as proof that by combining these methods, the retailing sector might come up with efficient inventory management strategies. Further work could also aim at incorporating more sources of external data into the model, and model scaling to allow real-time usage.

Download


Paper Citation


in Harvard Style

Bai Y. (2024). Machine Learning Implementation for Demand Forecasting in Supply Chain Management. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 77-84. DOI: 10.5220/0013206900004568


in Bibtex Style

@conference{ecai24,
author={Yumo Bai},
title={Machine Learning Implementation for Demand Forecasting in Supply Chain Management},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={77-84},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013206900004568},
isbn={978-989-758-726-9},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Machine Learning Implementation for Demand Forecasting in Supply Chain Management
SN - 978-989-758-726-9
AU - Bai Y.
PY - 2024
SP - 77
EP - 84
DO - 10.5220/0013206900004568
PB - SciTePress