The application of medoid-based cluster validation in desirable dietary pattern data

Abstract

A desirable dietary pattern (DDP) index is an index to measure the balance and variance of the nutrition intake of an individual. This index is composed of the calory values of protein, fat, and carbohydrates. Grouping individuals based on the DDP index is required to measure and improve an individual food security state. We took 14 individual purposively as samples to fill a set of DDP questioner. They were asked about their daily food consumption. They were grouped based on the DDP variables. A 3-dimensional plot showed that there were three to four clusters. Then, medoid-based partitioning algorithms, namely partitioning around medoids (PAM) and simple k-medoids (SKM), were applied in the data set. The inputted distances were generalized distance function to vary the distance options. The cluster results were then validated by medoid-based shadow value validation. This index was comparable to the 3-dimensional plot such that four clusters were opted as the most suitable number of clusters. The barplot of the cluster results showed that cluster 1 was characterized by an abundance of fat, while cluster 2 had very sufficient carbohydrates. Cluster 3 and 4 were two clusters with opposite characteristics where the former had a shortage of protein, fat, and carbohydrates, while the latter had an abundance of them.

Cluster Identification of Agrotourism in Banten Province Indonesia

“…Then, clustering results were validated via relative and internal criteria. The relative criterium applied a consensus matrix [18], while the internal criterium implemented silhouette width [19] and medoid-based shadow value [20][21]. The results from both criteria were visualized via bar-plot and spider-plot.…”