predictedPayments,AnnualAggLossDevModelOutput-method {BALD} | R Documentation |
A method to plot predicted vs actual payments for models from the BALD package.
object |
The object of type |
type |
A singe character value specifying whether to plot/return the predicted incremental or cumulative payments. Valid values are "incremental" or "cumulative." See details as to why these may not match up. |
logScale |
A logical value. If |
mergePredictedWithObserved |
A logical value. If |
plotObservedValues |
A logical value. If |
plotPredictedOnlyWhereObserved |
A logical value. See details. |
quantiles |
A vector of quantiles for the predicted payments to return. Usefull for constructing credible intervals. |
plot |
A logical value. If |
Because the model is Bayesian, each estimated payment comes as a distribution. The median of this distribution is used as a point estimate when plotting and/or returning values. Note: One cannot calculate the estimated incremental payments from the estimated cumulative payments (and vice versa) since the median of sums need not be equal to the sum of medians.
If mergePredictedWithObserved=TRUE
and type="incremental"
, then any observed incremental payment will be used in place of its corresponding incremental payment.
If mergePredictedWithObserved=TRUE
and type="cumulative"
, then only predicted incremental payments (by row) to the right of the last observed cumulative value will enter the calculation.
Mainly called for the side effect of plotting. Also returns a named array (with the same structure as the input triangle) containing the predicted log incremental payments. Returned invisibly.