I am currently a postdoc at Wake Forest University with Kenneth Berenhaut. I think about problems in applied discrete mathematics. Specifically, the questions I’m interested in arise from clustering, local depth and permutation-based techniques in time series analysis.
In joint work with Kenneth Barenhaut, we introduce a new clustering method that is completely non-parametric, does not require a Euclidean embedding and is based on a natural local depth function. A pre-print will be available shortly – feel free to contact me in the meantime if you would like more information. Above is the clustering of some two-dimensional Euclidean data obtained from this method.