Comparative Analysis of K-Means Variants Implemented in R

Almanza-Ortega, Nelva Nely; Pérez-Ortega, Joaquín; Dı́az, José; Solís-Romero, José

doi:10.13053/cys-26-1-4158

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…K-means clustering approach is used for the following reasons: (1) simplicity; (2) the data comes from survey responses treated as a ratio or continuous scale (Wu, Wang, & Wu, 2022) and (3) if the dataset is not large (Yuan & Yang, 2019). Hartigan-Wong K means employed since, compared to other available algorithms, it first allocates all data points to random centroids (Almanza-Ortega, Pérez-Ortega, Zavala-Díaz, & Solís-Romero, 2022).…”

Section: Methodsmentioning

confidence: 99%

Work from home connection: a cluster analysis based on the Internet service attributes towards subscribers profile

Paulino

Esteban²

2022

DTS

View full text Add to dashboard Cite

PurposeThe purpose of the study is to discover clusters or homogenous groups of work-from-home (WFH) Internet subscribers in the Philippines based on the attributes of speed, network quality, customer service, after-sales support, price, contract and value for money. Another objective is to determine if the formed clusters are related to the demographic profile.Design/methodology/approachA total of 275 internet subscribers from Metro Manila were surveyed. K-means cluster analysis using the Hartigan-Wong algorithm was performed on the data to generate the clusters.FindingsResults generated four significant clusters, which were named service value expecters, average expecters, low expecters and high expecters. Most of the subscribers are under the high expecters, followed by the service value expecters. The age and income of the subscribers are the profile that can affect the formulation of clusters in the Internet service industry. Those people in the younger age groups can be seen as more demanding, while older people tend to be content with the Internet service. Counter-intuitively, people with higher income seem to be more easily satisfied with Internet service features, while those people with lower income seem to be more demanding. Educational attainment and the number of household members do not have a direct effect on the formulation of clusters of Internet subscribers.Originality/valueThis is the first study to use cluster analysis in identifying possible segments of WFH Internet subscribers based on Internet service attributes. Its value is the provision of market segments based on which can be the basis for formulating marketing strategies and policies for the telecommunications industry.

show abstract

Section: Methodsmentioning

confidence: 99%