I just use the standard SPSS function to render cluster analysis and think / hope there are better ways to graph cluster analysis results.
I believe the default options in SPSS are to do a hierarchical, divisive clustering. If so, then the "dendrite" diagram that SPSS produces is the standard way of representing the results.
R software provides several packages that cover a significant variety of methods and tools in cluster analysis: “clustertend” to assess the cluster tendency in data, “fpc” for flexible clustering methods and cluster validation, “TSclust” for clustering time series (Montero & Vilar, 2014), “kml” and “kml3d” for longitudinal data clustering (Genolini et.al, 2015) or “clValid” for statistical and biological validation of clustering results (Brock et.al 2008), to name only a few. You can also opt for fuzzy clustering that can be implemented using the "fclust" package routines.
A different approach is to begin by running Multi-Dimensional Scaling on the same data, and determining whether the fit (stress) will accept a 2 dimensional solution. If so, you can map the clusters onto the MDS.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS