Skip to main navigation Skip to search Skip to main content

Comparative Analysis of Machine Learning and Deep Learning Models for Classification and Regression Tasks in Supply Chain Data

  • Atalay Tural*
  • , Bilal Saoud
  • , Ibraheem Shayea
  • , Abdulraqeb Alhammadi
  • , Jibran K. Yousafzai
  • *Corresponding author for this work
  • Istanbul Technical University
  • University of Mohamed El Bachir El Ibrahimi of Bordj Bou Arreridj
  • Astana IT University
  • Universiti Teknologi Malaysia
  • American University of Kuwait

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The increasing availability of large-scale supply chain data has created new opportunities for applying machine learning (ML) and deep learning (DL) techniques to support predictive analytics and operational planning. This study presents a comparative evaluation of ML and DL models using the DataCo SMART SUPPLY CHAIN FOR BIG DATA ANALYSIS dataset, which contains approximately 180,000 supply chain transactions. The analysis addresses both classification and regression tasks, including fraud detection, late-delivery prediction, sales forecasting, and order quantity prediction. Nine classification algorithms, seven regression models, and neural network architectures are implemented and evaluated using accuracy, recall, F1-score, mean absolute error (MAE), and root mean square error (RMSE). Experimental results show that the Decision Tree classifier achieved the best performance among the evaluated classification models, obtaining F1-scores of 81.0% for fraud detection and 99.42% for late-delivery prediction. For regression tasks, Linear Regression provided the most accurate sales pre dictions, while Random Forest and Extreme Gradient Boosting (XGBoost) achieved the best results for order quantity prediction. Neural network models demonstrated competitive performance, particularly in fraud detection, achieving a binary accuracy of 97.6%. The findings provide a comprehensive benchmark of ML and DL approaches for supply chain analytics and demonstrate the effectiveness of data-driven predictive models for supporting fraud detection, delivery management, and demand forecasting in supply chain environments.

Original languageEnglish
Title of host publication2026 International Conference on Cybersecurity, Digital Forensics, and AI Applications, ICCSDFAI 2026
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages293-297
Number of pages5
ISBN (Electronic)9798319520487
DOIs
Publication statusPublished - 2026
Event2026 International Conference on Cybersecurity, Digital Forensics, and AI Applications, ICCSDFAI 2026 - Istanbul, Turkey
Duration: 25 Jun 2026 → 27 Jun 2026

Publication series

Name2026 International Conference on Cybersecurity, Digital Forensics, and AI Applications, ICCSDFAI 2026

Conference

Conference2026 International Conference on Cybersecurity, Digital Forensics, and AI Applications, ICCSDFAI 2026
Country/TerritoryTurkey
CityIstanbul
Period25/06/26 → 27/06/26

Bibliographical note

Publisher Copyright:
© 2026 IEEE.

Keywords

  • Classification Models
  • Data Analysis
  • Deep Learning (DL)
  • Internet of Things (IoT)
  • Machine Learning (ML)
  • Regression Models
  • Supply Chain Systems

Fingerprint

Dive into the research topics of 'Comparative Analysis of Machine Learning and Deep Learning Models for Classification and Regression Tasks in Supply Chain Data'. Together they form a unique fingerprint.

Cite this