CUSTOMER CLASSIFICATION MODELING USING LDA CONSIDERING SPATIAL AREA STRUCTURE

Main Article Content

Hikari Ishihara
Takeo Okazaki

Abstract

To improve operational efficiency in the driver service industry, it is necessary to understand customer behavioural trends and characteristics. In this study, we propose a customer clustering method using customer data accumulated in a chauffeur service dispatch app, based on Latent Dirichlet Allocation (LDA), which incorporates spatial area structure as a feature. Specifically, we combine DBSCAN and GMM to zone areas with a high concentration of order locations and generate features that represent the area. Furthermore, we design a dataset that combines basic information such as customer age and time of use and extract potential customer groups using LDA. Finally, we evaluate the validity of the classification results and consider their applicability to service improvements

Downloads

Download data is not yet available.

Article Details

Section
Articles