Dr. Prakash Patil
Associate Professor of Statistics
Research interests: Nonparametric curve estimation, kernel- and wavelet-based nonparametric smoothing methods, robust statistical methods.
Author Profile on Google Scholar
Author Profile on MathSciNet
My current research interests are in goodness-of-fit tests and studying symmetry (or asymmetry) in functions. To test whether the given sample is consistent with the hypothesized theoretical probability distribution one compares frequency distribution over a fix number of classes with the expected frequencies of those classes under the theoretical probability distribution. If the underlying distribution is of continuous type it may make sense to allow the width of class intervals of the frequency distribution to go to zero as sample size increases (i.e. number of class intervals to grow with the sample size). Here main interest is to developing such procedures and study their properties in the context of goodness-of-fit for density and hazard rate functions.
Assumption of symmetry plays an important role in statistical inference and thus it is important to verify such assumption. Here my interest is first to quantify the size of asymmetry in a continuous probability distribution and then develop the procedures to test symmetry so that the size asymmetry is reflected in the power of those tests. In this direction in univariate set up we have achieved reasonable success and more focus is on the multivariate set up. In multivariate set up problem does become more interesting and involved because of many different kinds of symmetries (e.g. central symmetry, circular symmetry, elliptical symmetry etc) that one associate with a multivariate distribution.
Degree: Ph.D. in Statistics, 1990, Department of Statistics, University of North Carolina at Chapel Hill
Office Number: Allen 430