HT04_Lag: Implements the version of the conditional multivariate...

In rjaneUCF/MultiHazard: Tools for modeling compound events

Implements the version of the conditional multivariate approach of Heffernan and Tawn (2004) proposed in Keef et al. (2013) which incorporates lags between the variables.

Description

Implements the version of the conditional multivariate approach of Heffernan and Tawn (2004) proposed in Keef et al. (2013) which incorporates lags between the variables. Function utilizes the mexDependence and predict.mex.conditioned functions from the texmex package.

Usage

HT04_Lag(
  data_Detrend_Dependence_df,
  data_Detrend_Declustered_df,
  Lags,
  u_Dependence,
  Migpd,
  mu = 365.25,
  N = 100,
  Margins = "gumbel",
  V = 10,
  Maxit = 10000
)

Arguments

data_Detrend_Dependence_df	A data frame with (n+1) columns, containing in column 1 - Continuous sequence of dates spanning the first to the final time of any of the variables are recorded. 2:(n+1) - Values, detrended where necessary, of the variables to be modelled.
data_Detrend_Declustered_df	A data frame with (n+1) columns, containing in column 1 - Continuous sequence of dates spanning the first to the final time of any of the variables are recorded. 2:(n+1) - Declustered and if necessary detrended values of the variables to be modelled.
u_Dependence	Dependence quantile. Specifies the (sub-sample of) data to which the dependence model is fitted, that for which the conditioning variable exceeds the threshold associated with the prescribed quantile. Default is 0.7, thus the dependence parameters are estimated using the data with the highest 30% of values of the conditioning variables.
Migpd	An Migpd object, containing the parameterized Pareto models fitted (independently) to each of the variables.
Margins	Character vector specifying the form of margins to which the data are transformed for carrying out dependence estimation. Default is "gumbel", alternative is "laplace". Under Gumbel margins, the estimated parameters a and b describe only positive dependence, while c and d describe negative dependence in this case. For Laplace margins, only parameters a and b are estimated as these capture both positive and negative dependence.
V	See documentation for mexDependence.
Maxit	See documentation for mexDependence.
Lag	Matrix specifying the lags. The no lag i.e. 0 lag cases need to be specified. Row n denotes the lags applied to the variable in the nth column of data_Detrend_Dependence_df. Column n corresponds to the nth largest lag applied to any variable. NA. Default is matrix(c(0,1,0,NA),nrow=2,byrow = T), which corresponds to a lag of 1 being applied to variable in the first column of data_Detrend_Dependence_df and no lag being applied to the variable in the second column of data_Detrend_Dependence_df.

Value

List comprising the fitted HT04 models Models, proportion of the time each variable is most extreme, given at least one variable is extreme Prop, residuals z, as well as the simulated values on the transformed u.sim and original x.sim scales.

See Also

Dataframe_Combine Decluster GPD_Fit Migpd_Fit

Examples

HT04(data_Detrend_Dependence_df = S22.Detrend.df,
data_Detrend_Declustered_df = S22.Detrend.Declustered.df,
Migpd = S22_GPD, u_Dependence=0.7,Margins = "gumbel")

rjaneUCF/MultiHazard documentation built on April 9, 2024, 11:26 a.m.