A Visual Analytic System for Exploring Consumer Clusters

Ping Hsuan Huang, Yi Jheng Huang, Li Huang, Wen Chieh Lin

Research output: Contribution to journalArticlepeer-review

Abstract

Consumer transactions analysis is a fundamental component for companies to build strong customer relationships and make good decisions. Visualization can help with such tasks. Existing visualization methods of transaction data analysis often focus on specific purposes, such as abnormal behavior detection and stock analysis. Most of current systems focus on analyzing timevarying transaction pattern and usually focus on analyze webscrape data. Few of them are used to analyze the shopping behavior of customer clusters in physical stores.In this study, we present a visualization system to facilitate the process of transaction data exploration. Our system focuses on functions of customer clustering and exploration of customer characteristics. A distribution view embedded in our system visually demonstrates consumer clustering generated by a dimensional reduction algorithm. The visual clusters allow analysts to explore the characteristics of customers in different clusters. In addition, the correlation hinting method provided by our system automatically highlights overlapping subsets of consumers.It can guide analysts to explore interesting customer clusters. In sum, our system helps analysts find customers with similar behaviors, observe characteristics of interesting subsets, and determine the correlation among data attributes. We validate our system with the consumer transaction data from our collaborating department store. Use cases are provided to show the usability of the system.

Original languageEnglish
Pages (from-to)27-43
Number of pages17
JournalScientific Visualization
Volume13
Issue number1
DOIs
StatePublished - 2021

Keywords

  • Coordinated multiple view
  • Dimension reduction
  • Transaction behavior analysis
  • Visualization system

Fingerprint Dive into the research topics of 'A Visual Analytic System for Exploring Consumer Clusters'. Together they form a unique fingerprint.

Cite this