claimFitting {cascsim}

Claim data fitting analysis at line/type/status level

Description

Claim data fitting analysis at line/type/status level

Usage

claimFitting(object, claimData, ...)

## S4 method for signature 'Simulation,data.frame'
claimFitting(object, claimData,
  startDate = as.Date("2012-01-01"),
  evaluationDate = as.Date("2016-12-31"), lineList = object@lines,
  typeList = object@types, discreteDist = c("Poisson",
  "NegativeBinomial", "Geometric"), continuousDist = c("Normal",
  "Lognormal", "Pareto", "Weibull", "Gamma", "Uniform", "Exponential"),
  copulaList = c("normal"), fReportLag = TRUE, fSettlementLag = TRUE,
  fFrequency = TRUE, fSeverity = TRUE, fSSRCorrelation = TRUE,
  fFreqCorrelation = TRUE, copulaTest = TRUE, iTotalLoss = TRUE,
  fDeductible = TRUE, fLimit = TRUE, check = TRUE)

Arguments

object

Simulation object

claimData

claim data including existing claims for RBNER and claim reopenness analysis

...

Additional parameters that may or may not be used.

startDate

Date after which claims are analyzed;

evaluationDate

Date of evaluation for existing claims and IBNR;

lineList

List of business lines to be included in claim fitting;

typeList

List of claim types to be included in claim fitting;

discreteDist

List of discrete distributions to try fitting (report lag, settlemet lag, frequency);

continuousDist

List of continuous distribution to try fitting (severity);

copulaList

List of copula to try fitting;

fReportLag

Boolean variable to indicate whether report lag needs to be fitted;

fSettlementLag

Boolean variable to indicate whether settlement lag needs to be fitted;

fFrequency

Boolean variable to indicate whether monthly frequency needs to be fitted;

fSeverity

Boolean variable to indicate whether severity needs to be fitted;

fSSRCorrelation

Boolean variable to indicate whether copula among severity, report lag and settlement lag needs to be fitted;

fFreqCorrelation

Boolean variable to indicate whether copula among frequencies of business lines needs to be fitted.

copulaTest

Whether to test copula. The testing could take a very long time;

iTotalLoss

Boolean variable to indicate whether total loss before deductible and limit is available for severity fitting;

fDeductible

Boolean variable to indicate whether deductible empirical distribution needs to be fitted;

fLimit

Boolean variable to indicate whether limit empirical distribution needs to be fitted;

check

Boolean variable to indicate whether graph of each tried distribution fitting needs to be generated and saved.

Examples

library(cascsim)
data(claimdata)
lines<-c("Auto")
types<-c("N")
#exposure index
index1 <- new("Index",monthlyIndex=c(rep(1,11),cumprod(c(1,rep(1.5^(1/12),11))),
cumprod(c(1.5,rep((1.3/1.5)^(1/12),11))),
cumprod(c(1.3,rep((1.35/1.3)^(1/12),11))),cumprod(c(1.35,rep((1.4/1.35)^(1/12),11))),rep(1.4,301)))
#severity index
index2 <- new("Index",monthlyIndex=c(cumprod(c(1,rep(1.03^(1/12),59))),rep(1.03^(5),300)))
objan <- new("ClaimType", line="Auto",claimType="N",exposureIndex=index1,severityIndex=index2)
objlist <- list(objan)
simobj <- new("Simulation",lines=lines,types=types,claimobjs=objlist,iFit=TRUE,
iCopula=FALSE, iReport=TRUE, workingFolder=tempdir())
simobj <- claimFitting(simobj,claimdata,fSSRCorrelation = FALSE, fSettlementLag = FALSE)