|
NPStat — Non-parametric Statistical Modeling and Analysis
The NPStat package is designed to address the problem of non-parametric
statistical modeling of probability densities, Poisson intensity functions, and regression surfaces.
This type of modeling becomes very useful when there is little prior
information available to justify an assumption that the data belongs to
a certain parametric family of distributions or curves. The main
intended application is fast and detailed modeling of the response and
transfer functions of particle detectors, but the package is
sufficiently general and can be used for solving a variety of
statistical analysis problems from other areas. Both univariate and
multivariate models are supported, and a number of original algorithms
are implemented. Package capabilities include
- Calculation of descriptive sample statistics
- Arbitrary-dimensional histogramming
- Parametric, semi-parametric and non-parametric density modeling
- Generation of pseudo- and quasi-random numbers according to various density models
- Non-parametric density interpolation (morphing), including multivariate densities
- Non-parametric copula modeling and copula-based density interpolation
- Fast kernel density estimation (KDE) via DFFT
- Density estimation by local orthogonal polynomial expansion (LOrPE)
- Density estimation by the nearest neighbors method
- Local
regression techniques: local polynomial least squares, iterative local
least trimmed squares, local logistic regression, local quantile
regression with and without censoring
- Expectation-maximization unfolding with smoothing
- Edgeworth expansions of likelihood-based signal detection statistics
More information about NPStat statistical techniques and their implementation is available from this document.
The C++ API (doxygen)
Information about the python API
Download NPStat
Contact: Igor Volobouev
|