errorABC {poolABC} | R Documentation |
Calculate cross-validation prediction error of parameter estimation
Description
This function calculates the prediction error between estimates obtained after a leave-one-out cross validation for ABC and the true parameter values.
Usage
errorABC(true, estimated)
Arguments
true |
is a matrix where each row corresponds to the true parameter values of a given simulation and each column to a different parameter. These parameters where used as the pseudo-observed targets in the simulation study. |
estimated |
is a matrix with the estimated parameter values. Each row
corresponds to the estimate of the true parameter values present in the
corresponding row of the |
Details
The prediction error is calculated as sum((E-T)^2) / (nval * var(T))
,
where T is the true parameter value, E is the estimated parameter value, and
nval is the number of points where the true and predicted values were
compared.
Value
a numeric vector with the prediction error for each parameter. If column names are present in the input matrices, then this vector will also be named with the parameter names.