Check the confint method.
check_confint(object, ref, level = 0.95, which = "shape", method = "proflik")Arguments
- object
An object with class
"TVGEV"representing a stationary model.- ref
An object with class
"gev.fit"created by the functiongev.fitof the ismev package.- level
Confidence level.
- which
Not used yet.
- method
See
confint.TVGEV. For now, only"proflik"is possible.
Value
Nothing is returned: on the plot built by
ismev::prof.xi, two thick vertical lines are added, showing
the confidence limits found by confint.TVGEV with
method = "proflik". Each vertical line should cut
the profiled log-likelihood curve at one of its two intersections
with the horizontal line.
Details
Confidence intervals on the parameters of object are
computed using the confint method for the class
"TVGEV". The confidence interval for the shape parameter is
compared to that given by the ismev::prof.xi function.
See also
Examples
data(portpirie)
df <- portpirie
df <- within(df, date <- as.Date(sprintf("%4d-01-01", Year)))
fitNSGEV <- try(TVGEV(data = df,
response = "SeaLevel", date = "date"))
fitismev <- gev.fit(xdat = df$SeaLev)
#> $conv
#> [1] 0
#>
#> $nllh
#> [1] -4.339058
#>
#> $mle
#> [1] 3.87474692 0.19804120 -0.05008773
#>
#> $se
#> [1] 0.02793211 0.02024610 0.09825633
#>
check_confint(fitNSGEV, ref = fitismev)
#>
#>
#> o Finding CI for "mu_0"
#>
#> o 95%, lower bound: 3.82
#> Constraint check 0.0000000, -2.4183
#> 95%, upper bound: 3.93
#> Constraint & value -0.0000030, -2.4183
#>
#>
#> o Finding CI for "sigma_0"
#>
#> o 95%, lower bound: 0.16
#> Constraint check -0.0000000, -2.4183
#> 95%, upper bound: 0.24
#> Constraint & value -0.0000000, -2.4183
#>
#>
#> o Finding CI for "xi_0"
#>
#> o 95%, lower bound: -0.22
#> Constraint check -0.0000000, -2.4183
#> 95%, upper bound: 0.17
#> Constraint & value -0.0000000, -2.4183
#> If routine fails, try changing plotting interval
