Stock keeping unit segmentation based on unsupervised learning

Period: 01.01.2018

Supervisors:

Assistant professor 

Dr. sc. ing.
Ilya Jackson
Engineering Faculty
Ilya Jackson

Academic degree and current position in TSI: assistant professor, researcher at the research cluster “Data Analytics and Artificial Intelligence

Previous experience: 4 years of lecturing and research at the Transport and Telecommunication Institute (TSI)

Membership: Reviewer of the Hawaii International Conference on System Sciences (HICSS) and Transport Journal. Member of the scientific committee of the International Scientific Conference “Transbaltica 2021”. Member of a working group European Transport Research Review (ECTRI), thematic group “freight & logistics”.

Academic experience: More than 13 publications indexed by SCOPUS and WoS. Author of such courses as “Risk management in supply chains”, “Modeling Logistic and transport processes”, “Logistic systems and chains”.

Teaching at post- and graduate level: Risk management in supply chains (MSc in Transport and Logistics), Modeling Logistic and transport processes (MSc in Transport and Logistics), Logistic systems and chains (MSc in Transport and Logistics).

Participation in projects: participated as a researcher in several European projects including ALLIANCE (ID: 692426) and ePIcenter (ID: 861584).

Research Interests: Supply chain management, applied machine learning, simulation modelling, operations research and metaheuristics.

Supervised Doctoral, Master and Bachelor Theses (number): 1 master student completed master’s degree under my supervision.

Awards: Young Researcher Award was received at the 20th International Multi-Conference Reliability and Statistics in Transportation and Communication, 2020, Riga, Latvia. Awarded paper: “Neuroevolutionary approach to metamodel-based optimization in production and logistics”

Project Type: Applied

Main Challenge:

An average inventory system contains immense number of stock keeping units (SKUs). In general case, it is computationally impossible to consider each item individually and manage it under individual inventory policy. This project aims to develop the unsupervised learning approach for solving stock keeping units segmentation problem. Since the “ground truth” is not known, the research aims to compare such clustering algorithms as K-means, mean-shift and DBSCAN based only on the internal evaluation, thus, this research may be considered as descriptive cluster analysis.

Funding: VIAA scholarship for doctoral research decision Nr.1.-50.3/3889

DA&AI tools:

  • Clustering (k-means, mean-shift, DBSCAN)
  • Principal component analysis
  • Anomaly detection (local outlier factor)
  • Missing values imputation (KNN imputation)

Deliverables (publications, reports, acknowledgments, etc.)

  • Jackson, I., Avdeikins, A. and Tolujevs, J. (2018) Unsupervised Learning-Based Stock Keeping Units Segmentation. In: proceedings of International Conference on Reliability and Statistics in Transportation and Communication. Springer, Cham, pp. 603-612 [indexed by Scopus].

contact us