MMU - MSc Multivariate Statistics, Cluster Analysis using Minitab

February weather example

This example uses the February weather example data file (download from the data files link on the unit WebCT homepage).

Step 1

Open the worksheet (NOT a project - by default Minitab will attempt to open a project)

Minitab File Menu with Open Worksheet highlighted

Note that you may have to navigate to the correct file location using the 'Look in' down arrow on the Open Worksheet window.

Windows file selection window

Step 2

Select the correct cluster observations option and then variables to use for the clustering. This example clusters observations (the years).

Minitab Stat>Multivariate dropdown menu with Cluster Observations highlighted

All of the variables, except year, should be selected. Year is not used in the clustering because we want to examine the membership of cluster with respect to the year.

Part of the Minitab Cluster Observations window with all varaibles except YEAR selected

Step 3

Select the cluster options: Linkage measure and Distance Measure. In this example Single Linkage clustering (nearest neighbour) has been combined with a Euclidean distance measure.

Part of the Cluster Observations window with Single Linkage selected as the method

Part of the Cluster Observations window with Euclidean selected as the distance measure

Next, decide if you want to standardise the variables (not standardised in this example) and have the dendrogram drawn. If you don't tick this option you won't get a dendrogram.

Part of the Cluster Observations window with the Show Dendrogram oprion ticked

The Customize button can be used to customize the appearance of the dendrogram. In this example the variable YEAR has been selected to label the cases.

Part of the Cluster Observations customise window with YEAR selected as the case labels

The result is a dendrogram