Determination of Hidden Extrapolations via Gaussian Mixture Models

Abstract

Testing and experimentation can, in general, be a costly or time consuming endeavor and, as such, companies are motivated to minimize costs. This often causes engineers to merge many different data sets, resulting in imbalanced multidimensional spaces of data. As a consequence, it becomes increasingly difficult to locate where hidden extrapolations have occurred when using these amorphous data sets. Thus, there is a need for a more rigorous method in extrapolation determination without requiring engineers to spend large of amounts of time parsing multidimensional data. A method relying on Gaussian Mixture Models for hidden extrapolation
determination is presented in this paper.