In this article
Once variables have been added and any adjustments to the advanced settings have been made, click Run cluster analysis to produce a calculation output as in the example below .
Figure 1 - Example of a Cluster Analysis output
Several variations of the cluster analysis can be made, with each calculation run saved separately within the cluster analysis object. The most recent calculation run will display first. Calculation runs can be renamed or deleted if required.
Click the Send to Excel button to export the tables to Excel.
The Variable Mean Values Table for Cluster Analysis
The primary output from the cluster analysis is a table displaying the proposed clusters, with a summary of the mean scores for each variable used per identified cluster . This information allows the analyst to understand the characteristics of each cluster and provide insights into identifying and labeling the unique cluster segments.
Figure 2 - The Variable mean values table
Looking at the results from the example data, we see that there are some defining values in the clusters which we may use to segment our customer base.
- The first cluster contains individuals who have a relatively high product score, and a high loyalty score. We may say that for these people, the quality of the product has a high impact on their loyalty (for example product advocates).
- The second cluster contains individuals that have lower product and loyalty scores, but provide a high rating for service. For these customers, we can say that the quality of the ongoing service is a key element for keeping them loyal.
- In the third cluster, there are low scores for both loyalty and service, despite a relatively high score for the product itself. The customers in this segment are at risk of switching products due to poor service, despite valuing the product itself.
PCA Chart
If the option to “Show PCA chart” is selected in the advanced properties, then a scatter plot is created to visually illustrate the cluster groupings . Where there are more than two variables used in the cluster analysis, the variables are reduced into the two primary principal components in order to create a two-dimensional chart, whilst retaining as much information about the underlying variables (variance) as possible. The total amount of variance accounted for by the principal components is provided beneath the chart.
Figure 3 - The PCA Chart