CMStatistics 2018: Start Registration
View Submission - CMStatistics
B1066
Title: Computational tools and methods for statistical inference based on using the characteristic functions Authors:  Viktor Witkovsky - Slovak Academy of Sciences (Slovakia) [presenting]
Abstract: The exact distributions of many estimators and test statistics can be specified by their characteristic functions. Frequently, such distributions can be expressed as linear combinations or products of independent random variables with known parametric and/or non-parametric (empirical) distributions, specified by their characteristic functions. We consider computational methods and algorithms for specific problems (as e.g. the exact tests in multivariate analysis or non-parametric methods) of statistical inference based on combining characteristic functions and their numerical inversion to evaluate the associated CDF/PDF and quantiles. New methods and algorithms are being continuously developed and implemented into the MATLAB resp. R version of CharFunTool - the Characteristic Functions Toolbox. We shall present current status and basic functionality of the toolbox and illustrate its usage for selected problems of statistical inference.