errint (version 1.0)

acc_intervals: Accuracy of Error intervals

Description

int_intervals computes the real accuracy of a given error intervals for a particular set of errors and a particular error function.

Usage

acc_intervals(interv, errors, f = function(x, y) {     abs(x - y) }, tol = 10^-8)

Arguments

interv

error interval.

errors

set of errors.

f

error function to be used to compute error between real x (interv) and predicted y (errors) values. See also 'Details'.

tol

used to normalize residual values to (0,1) when beta is the assumed distribution. See also 'Details'.

Value

Returns an object of class c("measure","list") with information of the interval accuracy.

Details

f must be a function that takes two arguments, x and y, and return a numeric value.

The formula used to normalize residual values to (0,1) when a Beta distribution is assumed is $|\phi|*(max(|\phi|)+tol)^(-1)$.

References

Link to the scientific paper

Prada, Jesus, and Jose Ramon Dorronsoro. "SVRs and Uncertainty Estimates in Wind Energy Prediction." Advances in Computational Intelligence. Springer International Publishing, 2015. 564-577,

with theoretical background for this package is provided below.

http://link.springer.com/chapter/10.1007/978-3-319-19222-2_47

See Also

measure error_interval

Examples

interv<-int_gau(rnorm(10),0.1)
acc_intervals(interv,rnorm(10))
acc_intervals(interv,rnorm(10),function(x,y){x-y})