A1203
Title: Hypothesis testing of mean interval for p-dimensional interval-valued data
Authors: Anuradha Roy - The University of Texas at San Antonio (United States) [presenting]
Fernando Montes - The University of Texas at San Antonio (United States)
Abstract: A new parametric hypothesis test of the mean interval for p-dimensional interval-valued (hyper-rectangles) dataset is proposed under the assumption that the lower bound and the upper bound of an interval are two repeated measurements and the p-dimensional lower bounds and p-dimensional upper bounds have the same variance-covariance matrix. An orthogonal transformation is employed to obtain an equivalent hypothesis test of p-dimensional mean interval of interval-valued dataset in terms of a normal p-dimensional vector of midpoints and a lognormal p-dimensional vector of ranges of the p-dimensional interval-valued dataset. The mean vector of the normal data is tested using Hotellings T-squared test, while testing for the mean vector of the lognormal data is performed via the construction of a generalized pivotal quantity in a Monte Carlo simulation. Independence of the two tests allows the use of the Benjamini-Hochberg multiple testing procedure to control the false discovery rate. Simulations are conducted to compute the empirical Type I errors of the test statistic at nominal significance levels 0.01 and 0.05. The performance of the proposed test is illustrated with real-life examples.
