From Disorder to Structure: Hierarchical Analysis of Market Basket Data
Discover Economics and Finance · 07.08.2025 · 3 (2) · 01-04
DOI: 10.51470/def.2025.3.2.01 ↗Kurz gefasst
Fachartikel in der Zeitschrift Discover Economics and Finance, online erschienen am 11.09.2025. Verfasst von Pala und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Market basket data is inherently disordered: transactions arrive as unordered sets of items with no explicit relational labels. This paper presents a hierarchical clustering framework that recovers latent categorical structure directly from co-purchase statistics, without requiring a predefined product taxonomy. We formalize a Jaccard-similarity-based distance measure over items, apply agglomerative hierarchical clustering, and evaluate the fidelity of the resulting dendrogram using cophenetic correlation and cluster purity against a known ground-truth grouping. On a synthetic but realistically structured retail transaction log (3929 transactions, 22 items across six product categories), the method achieves a cophenetic correlation of 0.9973 and cluster purity of 1.0000, indicating that hierarchical clustering recovers the underlying category structure with high fidelity. We discuss implications for automated category management, cross-selling, and store layout optimization, and outline limitations that arise when co-purchase signals are sparse or noisy.
Zitieren
APA
Pala, Maqbool, M., & Lone, A. A. (2025). From Disorder to Structure: Hierarchical Analysis of Market Basket Data. Discover Economics and Finance, 3(2), 01-04. https://doi.org/10.51470/def.2025.3.2.01
Vancouver
Pala, Maqbool M, Lone AA. From Disorder to Structure: Hierarchical Analysis of Market Basket Data. Discover Economics and Finance. 2025;3(2):01-04. doi:10.51470/def.2025.3.2.01
BibTeX
@article{Pala2025,
author = {Pala and Maqbool, M and Lone, Ajaz Ahmad},
title = {From Disorder to Structure: Hierarchical Analysis of Market Basket Data},
journal = {Discover Economics and Finance},
year = {2025},
volume = {3},
number = {2},
pages = {01--04},
doi = {10.51470/def.2025.3.2.01}
}
EUVIMED