Single Variable Cdf Estimator

A Single Variable Cdf Estimator (more precisely a conditional-cdf estimator) estimates a conditional distribution function of variable Z at a location u given some neighboring Z-values. It can estimate parametric cdf's (usually Gaussian) or non-parametric cdf's, using different approaches: kriging, search trees (SNESIM) or even neural networks.

Refinement of

Assignable

Notations

a		an object of a type that models Cdf Estimator
f		an object of a type that models CDF
u		an object of a type that models a Location
V		an object of a type that models a Neighborhood

Valid Expressions

• Estimate cdf

  a(u,V,f)

  Return type:		int
  Semantics:		estimates the ``parameters'' of cdf f (mean and variance in the case of a Gaussian cdf, or the probabilities Prob(Z$\le$zi) = F(zi), i = 1,..., n, if F is non-parametric). The returned value is 0 if no problem was encountered during the execution of the function. If the returned value is different from 0 (i.e. the estimation of f failed), ccdf f has not been modified by the function call

Models

• gaussian_cdf_Kestimator: estimator of a gaussian cdf using kriging (K)
• indicator_cdf_Kestimator: estimator of a non-parametric cdf using indicator kriging
• search_tree_estimator: estimator of a non-parametric cdf using a search tree (SNESIM)