Biometrics Northwest LLC

Performing Statistical Data Analysis and Modeling



About Us


Contact Us


Project: Nonparametric Probability Density Estimation

A MATLAB toolbox 'bsspdfest' implementing nonparametric probability function estimation using normalized B-splines was developed. The toolbox implements nonparametric probability function estimation procedures for one or more dimensions using a B-spline series for one-dimensional data and a tensor product B-spline series for multi-dimensional data. The toolbox takes advantage of the direct addressing of MATLAB arrays up to three dimensions and various vectorization approaches to speed up the computations. For data dimensions greater than three indirect addressing is used, converting multi-dimensional indices into linear array addressing, making this function slower.

The toolbox supports estimation and computation of the PDF, CDF, and survivor functions for data of all dimensions as well as the inverse CDF (ICDF) and cumulative hazard functions for one dimensional data. The toolbox also supports the creation and use of gridded interpolants to provide very fast approximate evaluation of the B-spline series or tensor product series for the probability functions. Bounded domains are also supported for all dimensions.

Version 3.1.0 of the bsspdfest toolbox has just been released! The major changes were to the interfaces for the estimation and evaluation functions. They were changed from using positional arguments to varargin. In addition, a memory saving evaluation of gridded interpolants was introduced to avoid generating the entire grid of points before creating the gridded interpolant. A variety of minor improvements and corrections to the documentation in the headers were also made. See the Changes.txt file for details.

A mixture of 5 2-D normal distributions was used as an example. The data, the estimated and true PDFs, and the error are shown below.


Estimated and true PDFs


Two three dimensional visualizations were generated using a mixture of five multivariate normal distributions. In addition, a more whimsical data set in the shape of a cat was created and the bsspdfest toolbox was used to estimate a probability density function for the data-cat, producing a proba-bili-densi-cat and a data-cat-scan visulization.
3d mixture distrbution x3d mixture distrbution z Data-cat scan

The MATLAB toolbox, a user's guide, and examples are available in the ZIP file Be sure to read the license and see what changes were made for the current version.

Back to top

For information send email to:

Last Update: December 23, 2016 12:56 PM

Copyright 2005-2016 Biometrics Northwest LLC