Cluster Analysis featured in Optimus 10.14 takes post-processing analysis to a higher level. By grouping design points with similar characteristics in separate clusters, Optimus is able to identify valuable correlations between and within clusters. As the new Optimus clustering capability is fully automated, it combines unmatched ease of use and high consistency in delivering accurate results. In addition, Optimus supports visualization tools for easy graphic cluster evaluation including cluster scatter and parallel coordinates charts.
Engineers often use cluster analysis to identify interesting subregions in the design space to be explored in greater detail. Clustering may also trace correlations between designs within the same cluster, which may not be visible from an analysis covering all designs. Besides gaining deeper insight through valuable data correlations, Cluster Analysis can save tremendous simulation time. In specific situations, the dataset can be reduced reliably with individual cluster-representative data points instead of using all cluster points.
Next to K-means and Hierarchical clustering methods, Optimus incorporates Gaussian Mixture Models (GMM) clustering. This extremely powerful clustering method is ideal for large datasets with varying cluster sizes. Optimus sets itself apart by automatically calculating the appropriate number of clusters for the specific application under investigation. Identifying the most suitable cluster count is critical in obtaining an automated cluster analysis process that consistently delivers accurate clustering results with minimum user interaction.