Using the Superformula, advances in n-dimensional data analyses are showing that current approaches can be improved substantially.
One of the limitations of current methods is the assumption of isotropic n-dimensional space. This results in a failure rate that can easily rise above 30%, when missing data points from the collection that they should belong to, or when data points get included where they should not be. Another limitation is that groups or subgroups might go undetected at all. Better approaches use perceptron networks. They can be tuned quite delicately, but need training.
Still, the results in a head-to-head comparison showed a failure rate of approximately 3%.
This has been the most accurate approach so far.
Using the ability to find non-isotropical collections of data without the need of training, the accuracy could be approved. By applying the Superformula to n-dimensional datasets, a failure rate of only 1,7% was shown, which is nearly cutting the currently best achievable failure rate in half.Applications vary across various fields, such as marketing, forensic research, new product development, and big data analytics.