Introduction

• This is a very simple demonstration of k-means clustering under shiny.
• Dataset is all.mammals.milk.1956 from R package ‘cluster.datasets’.
• The dataset contains a list of animals and the constituents of their milk.
• The dataset have 25 observations on 6 variables.
• Data Source: http://people.sc.fsu.edu/~jburkardt/datasets/spaeth2/spaeth2.html

Here is a look of the dataset:

##        name water protein fat lactose  ash
## 1     Horse  90.1     2.6 1.0     6.9 0.35
## 2 Orangutan  88.5     1.4 3.5     6.0 0.24
## 3    Monkey  88.4     2.2 2.7     6.4 0.18

Code ui

• It has a headerPanel for the title
• It has a sidebarPanel with numericInput for the Cluster Count (range from 1 to 9)
• sidebarPanel outputs k-Means calculated centroids coordinates.
• mainPanel plots all points with the centroids shown in real time right after the Cluster Count changes.
• Code for getting the Cluster Count Input :

numericInput('clusters', 'Cluster count', 1, min = 1, max = 9)

Code server

• It loads the R package “cluster.datasets”
• Then import the dataset “all.mammals.milk.1956”
• It gets the ui Cluster Count input-“cluster”
• Then calculate the centroids coordinates using kmeans function
• Finally it render the plots as well as the centroid coordinates table for the ui to display
• Code for the kmeans evaluation:

clusters <- reactive({
       kmeans(selectedData(), input$clusters)

Published on Shiny