Last data update: 2014.03.03

R: Compute return levels and confidence limits for a "Renouv"...
predict.RenouvR Documentation

Compute return levels and confidence limits for a "Renouv" object

Description

Compute return levels and confidence limits for an object of class "Renouv".

Usage

   ## S3 method for class 'Renouv'
predict(object,
        newdata = c(10, 20, 50, 100, 200, 500, 1000),
        cov.rate = TRUE,
        level = c(0.95, 0.7),
        prob = FALSE,
        trace = 1, eps = 1e-06,
        ...)

Arguments

object

An object of class "Renouv" typically created by using the Renouv function.

newdata

The return period at which return levels and confidence bounds are wanted.

cov.rate

If FALSE, the delta method will not take into account the uncertainty on the event rate lambda of the Poisson process. Note however that when distname.y is "exponential" and when no MAX or OTS data is used, the value of cov.rate has no impact for now, because the delta method is not used then.

level

Confidence levels as in other 'predict' methods (not percentages).

prob

If TRUE a prob column is found in the returned data frame. This column can be used to find which quantile was used to compute the return level.

trace

Some details are printed when trace is not zero.

eps

Level of perturbation used to compute the numerical derivatives in the delta method.

...

Further arguments passed to or from other methods.

Details

Unless in some very special cases, the confidence limits are approximated ones computed by using the delta method with numerical derivatives.

Value

A data frame with the expected return levels (col. named "quant") at the given return periods, and confidence limits. The returned object has an infer.method attribute describing the method used to compute the confidence limits.

Note

Despite of its name, this method does not compute true predictions. A return period is to be interpreted as an average interevent time rather than the duration of a specific period of time. For instance, the expected return level for a given return period with length 100 years is the level that would be on average exceeded once every 100 years (assuming that the model description in object is correct).

Author(s)

Yves Deville

References

Coles S. (2001) Introduction to Statistical Modelling of Extremes Values, Springer.

See Also

Renouv to fit Renouv model.

Examples

## Use Brest data
fit <- Renouv(Brest)
pred <- predict(fit, newdata = c(100, 125, 150, 175, 200),
                level = c(0.99, 0.95))

Results


R version 3.3.1 (2016-06-21) -- "Bug in Your Hair"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-pc-linux-gnu (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.

> library(Renext)
Loading required package: evd
> png(filename="/home/ddbj/snapshot/RGM3/R_CC/result/Renext/predict.Renouv.Rd_%03d_medium.png", width=480, height=480)
> ### Name: predict.Renouv
> ### Title: Compute return levels and confidence limits for a "Renouv"
> ###   object
> ### Aliases: predict.Renouv
> 
> ### ** Examples
> 
> ## Use Brest data
> fit <- Renouv(Brest)
Special inference for the exponential case without history
Warning message:
In predict.Renouv(object = res, newdata = rl.period, level = round(pct.conf)/100,  :
  uncertainty on the rate not taken into account yet  in the exponential with no history case
> pred <- predict(fit, newdata = c(100, 125, 150, 175, 200),
+                 level = c(0.99, 0.95))
Special inference for the exponential case without history
Warning message:
In predict.Renouv(fit, newdata = c(100, 125, 150, 175, 200), level = c(0.99,  :
  uncertainty on the rate not taken into account yet  in the exponential with no history case
> 
> 
> 
> 
> 
> dev.off()
null device 
          1 
>