Welcome to the Project "Risk Assessment with R" !

This R-Forge project was first developed to help the administrators Marie-Laure Delignette Muller, Jean-Baptiste Denis and Regis Pouillot to share elements from the three R packages they are building, "fitdistrplus", "mc2d" and "rebastaba".

These packages are developed in the aim to help risk assessors to build and study their models using Monte-Carlo simulations with R.

fitdistrplus: Specification of a Parametric Distribution from Data

“fitdistrplus” is a set of functions specifically written:

to help to choose the best parametric distribution that fits a given dataset;
for a given distribution, to estimate the parameters using the maximum likelihood method or the method of matching moments and provide graphs and statistics to assess the fit. Specific functions were written for censored data (right- left- and interval-censored);
for a fitted distribution, to simulate the uncertainty in the estimated parameters by parametric or nonparametric bootstrap resampling.

A stable version is available on CRAN here!

"mc2d" is a set of functions specially written to:

provide additional distributions: Bernoulli, triangular, PERT, discrete empirical, generalized beta, (vectorized) multi-normal, (vectorized) Dirichlet, (vectorized) multinomial;
provide tools to build correlation structure between parameters: using the Iman & Conover (1982) method, possibly on more than two parameters;
provide tools to build models: the package transfers easily variability and uncertainty along a mathematical model;
provide tools to study the model: through ad-hoc summaries and graphics, sensitivity analyses (rank correlation on the variability or the uncertainty dimension), etc.

A stable version is available on CRAN here!

"rebastaba" is a set of functions specially written to manipulate Bayesian networks with a maximum of possibilities, among them :

nodes can be discrete or continuous variates,
nodes can be uni or multivariates,
the distribution of the node can be parametric or not (drawn from data.frames),
data sets simulated with the /bn/ can be obtained,
interactive graphical representation of the graph is available,
standard functions (non specific to /bn/s) are gathered in a sub-package named rs00,
...

The final aim is to provide connections with other softwares among them Jags.

For the moment, rebastaba is denominated as rbsb.

Feel free to join this project to: