[R-pkgs] Introducing empirical: Probability Distributions as Models of Data
I would like to introduce my R package:
empirical: Probability Distributions as Models of Data
The description is:
Computes continuous (not step) empirical (and nonparametric) probability
density, cumulative distribution and quantile functions. Supports
univariate, multivariate and conditional probability distributions, some
kernel smoothing features and weighted data (possibly useful mixed with
fuzzy clustering). Can compute multivariate and conditional probabilities.
Also, can compute conditional medians, quantiles and modes.
(1) I'm planning to support categorical variables in the future.
(2) There are some problems with univariate models (but not multivariate),
(3) Contrary to what the name empirical suggests, currently multivariate
models use kernel smoothing.
(4) I'm interested in implementing a hybrid Kernel-Quantile method, which I
suspect may be more robust to outliers. If I succeed, then I may rewrite
the univariate implementation.