This is an old revision of the document!
Cluster Calculations
This tool can be used if you have a big unlabeled dataset and you need to group all information into smaller categories or clusters. A number of clusters (k-centroids) needs to be defined for running the tool. It's an iterative process where each cluster is associated with a centroid. In the calculative process the sum of distances minimizes between the datapoints and their corresponding clusters. The tool will assign each datapoint to its closest k-centered centroid and a cluster is created hence each cluster will have commonalitites away from other clusters.
Alt om Cluster Wizard. The cluster calculation can now be run within and region. Furthermore, the output cluster points are shown on the top plot as default.
Requirements:
- 1D data
Step 1.
1. Open profile and get a data overview in order to evalute the number of clusters needed.
- A profile showing 1D PACES data.
1. “Themes” menu –> “Cluster 1D Models…”.
