Sparse hierarchical clustering based on Menopause Rating Scale severity of symptoms collected from perimenopausal and postmenopausal US women in a menopause tablet perceptual efficacy study

Abstract

AbstractThe sparse hierarchical clustering (SHC) method and the R package “sparcl” are used for consumer segmentation with 247 perimenopausal and postmenopausal US women. The data includes 11 menopause rating scale (MRS) attributes measuring severity of symptoms collected in the first week of a 12‐week extended use menopause tablet perceptual efficacy study. SHC is one of the advanced statistical clustering approaches for high‐dimensional data with potential noise variables. Sparse clustering is an important topic of the actively developing field of statistical learning with sparsity. An advantage of the SHC method is that it simultaneously finds the clusters of subjects and the important clustering variables. It allows clustering the data in low‐dimensional subspaces without dimension reduction of the data before clustering. The SHC method is particularly applicable to consumer segmentation based on high‐dimensional consumer data.Practical ApplicationsSparse hierarchical clustering (SHC) provides advantages in consumer segmentation based on high‐dimensional consumer data with potential noise variables. This advanced clustering method finds the relative importance of attributes in clustering, helping to identify homogeneous clusters and enhance understanding of how consumer responses/needs vary among clusters.