Usage

In SyScSelection: Systematic Scenario Selection for Stress Testing

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(SyScSelection)

Example ellipsodial mesh for a normal distribution:

• Estimate the mean and covariance matrix from the data:
  mu <- colMeans(data)
sig <- cov(data)

• The number of dimensions, d, is taken directly from the data:
  d <- length(data[1,])

• Get the size parameter for a normal dist’n at a 95% threshold:
  calpha <- sizeparam_normal_distn(.95, d)

• Create a hyperellipsoid object. Note that the constructor takes the inverse of the disperion matrix:
  hellip <- hyperellipsoid(mu, solve(sig), calpha)

• Scenarios are calculated as a mesh of fineness 3. The number of scenarios is a function of the dimensionality of the hyperellipsoid and the fineness of the mesh:
  scenarios <- hypercube_mesh(3, hellip)

Example ellipsodial mesh for a t distribution:

• Estimate the mean, covariance, and degrees of freedom from the data:
  mu <- colMeans(data)
sig <- cov(data)
nu <- dim(data)[1] - 1

• The number of dimensions, d, is taken directly from the data:
  d <- length(data[1,])

• Get the size parameter for a normal dist’n at a 95% threshold:
  calpha <- sizeparam_t_distn(.95, d, nu)

• Create a hyperellipsoid object. Note that the constructor takes the inverse of the disperion matrix:
  hellip <- hyperellipsoid(mu, solve(sig), calpha)

• Scenarios are calculated as a mesh of fineness 3. The number of scenarios is a function of the dimensionality of the hyperellipsoid and the fineness of the mesh:
  scenarios <- hypercube_mesh(3, hellip)

SyScSelection documentation built on Oct. 26, 2020, 5:08 p.m.