My latest paper The QFlex Distribution, coauthored with Professors J. Eric Bickel and Benjamin D. Leibowicz, is out as preprint on SSRN! In this paper, we introduce a new framework for constructing quantile-parameterized distributions from an arbitrary set of quantile assessments. At a high level, our approach takes a finite collection of point assessments from an unknown distribution and produces a high-fidelity estimate of the underlying distribution under very mild assumptions. This system enables decision makers to build meaningful probabilistic models from sparse expert assessments.

IISE Award Presentation
Figure. The QFlex system can match a wide variety of distribution shapes and boundedness. Here is it creating density functions from quantiles sampled from a set of bounded distributions.