# ELT - A package to build Experience Life Tables Collection of functions that can be used following a pre-established procedure to build and validate actuarial life tables. package ## Details ||| |:--|:--| |Package:|ELT| |Type:|Package| |Version:|1.6| |Date:|2016-04-10| |License:|GNU| |Depends:|locfit,lattice,latticeExtra,xlsx| The package is meant to be used following a pre-established procedure. See the reference for more info. Please notice that the package includes the following internal functions: .BeforeAfterCompletion(); .ComparisonFitsMethods(); .ComparisonFitsMethodsLog(); .ComparisonResidualsMethods(); .ComparisonResidualsMethods(); .ComparisonTrendsMethods(); .CompletionDG2005(); .CompLevel1(); .CompLevel2(); .CompLevel3(); .DevFct(); .ExportHistoryInExcel(); .ExportPeriodicLifeExpinExcel(); .ExportSingleIndiciesinExcel(); .ExportValidationL1inExcel(); .ExportValidationL2inExcel(); .FctCohortLifeExp5(); .FctPerLifeExp(); .FctSingleIndices(); .FitPopsAfterCompletionLog(); .FittedDxtAndConfInt(); .GetCritLevel1(); .GetCritLevel2(); .GetCV(); .GetFitSim(); .GetHistory(); .GetQtiles(); .GetRelDisp(); .GetSimExp(); .PlotCrit(); .PlotCritChoice(); .PlotDIntConf(); .PlotExpQtle(); .PlotFittedYear(); .PlotFittedYearLog(); .PlotMethod(); .PlotParamCompletion(); .PlotPerExp(); .PlotRelDisp(); .PlotRes(); .ResFct(); .SimDxt(); .ValidationLevel3(); .WarningInvalidAge() . These functions can be accessed with the prefix ELT::: using the following syntax: ELT:::[name of the function] . For example : ELT:::.GetHistory(). See technical note II1291-15 (http://www.ressources-actuarielles.net/gtmortalite) for the arguments and examples of the functions. ## References Tomas, J. , Planchet, F. , Prospective mortality tables and portfolioexperience, Chapter 9 in _Computational Actuarial Science_, with R ;Arthur Charpentier Editor, Chapman, 2014 Tomas, J. , Planchet, F. , Constructing entity specific prospectivemortality table : adjustment to a reference, _Les cahiers de recherchede l'ISFA_, 2013(13), pp.1-31, 2013. Tomas, J. , Planchet, F. , Construction d'une table de mortalite parpositionnement : Mode d'emploi, _Institut des Actuaires, Rapporttechnique II1291-15_, pp. 1-27, 2013 Tomas, J. , Planchet, F. , Criteres de Validation : AspectsMethodologiques, _Institut des Actuaires, Rapport technique II1291-14_,pp. 1-31, 2013 Tomas, J. , Planchet, F. , Methodes de positionnement : AspectsMethodologiques, _Institut des Actuaires, Rapport technique II1291-12_,pp. 1-12, 2013 Denuit, M. and Goderniaux, A. C. (2005). Closing and projecting lifetables using log-linear models. _Bulletin of the Swiss Association ofActuaries_, (1), 29-48 \ for data andexemple codes. ## Examples ```r ## Not run: data(MyPortfolio) data(ReferenceMale) data(ReferenceFemale) ## ------------------------------------------------------------------------ ## ## Initialize Age variables ## ## ------------------------------------------------------------------------ ## AgeRange <- 30:90 AgeCrit <- 30:90 AgeRef <- 30:95 History <- ReadHistory(MyPortfolio = MyPortfolio, DateBegObs = "1996/01/01", DateEndObs = "2007/12/31", DateFormat = " MyData <- AddReference(History = History, ReferenceMale = ReferenceMale, ReferenceFemale = ReferenceFemale) ## ######################################################################## ## ## METHOD 1 ############################################################### ## ## ######################################################################## ## ## ------------------------------------------------------------------------ ## ## Execute method 1 ## ## ------------------------------------------------------------------------ ## OutputMethod1 <- Method1(MyData = MyData, AgeRange = AgeRange, Plot = T) ## ------------------------------------------------------------------------ ## ## Validate method 1 by the 1st level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 1st level citeria. ValidationLevel1Method1 <- ValidationLevel1(OutputMethod = OutputMethod1, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Plot = T, Excel = T) ## ---------- If the criterions corresponding to the 1st level are not ## ---------- satisfied, we can modify the age range used to compute the SMR ## ---------- and reexecute ## ---------- OutputMethod1 <- Method1(...) ## ---------- and ## ---------- ValidationLevel1Method1 <- ValidationLevel1(...). ## ---------- If the criterions corresponding to the 1st level are still not ## ---------- satisfied, we turn to the method 2, and it is useless to ## ---------- pursue the completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the criterions corresponding to the 2nd level. ## ---------- We can also turn to method 3 or 4 to improve the fit at a cost ## ---------- of a somewhat greeter complexity. ## ------------------------------------------------------------------------ ## ## Validate method 1 by the 2nd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 2nd level criterions ValidationLevel2Method1 <- ValidationLevel2(OutputMethod = OutputMethod1, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Excel = T) ## ---------- If the criterions corresponding to the 2nd level are not satisfied ## ---------- we turn to the method 2 and it is useless to pursue the ## ---------- completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the completion of the table and the criterions corresponding to ## ---------- the 3rd level. ## ------------------------------------------------------------------------ ## ## Completion Method 1 ## ## ------------------------------------------------------------------------ ## ## ---------- Age range for the selection of the optimal starting age. AgeRangeOptMale <- AgeRangeOptFemale <- c(80, 80) ## ---------- In theory, we could select the optimal starting age, however ## ---------- the optimal starting age can vary a lot with the calendar years ## ---------- leading to a relatively irregular surface. In practice, we ## ---------- select then a fixed age for the whole years. ## ---------- Starting age for which the fitted probabilities of the death are ## ---------- replaced by the values obtained from the completion model. BegAgeCompMale <- BegAgeCompFemale <- 85 ## ---------- We check if the completion is smoothed with graphical ## ---------- diagnostics. CompletionMethod1 <- CompletionA(OutputMethod = OutputMethod1, MyData = MyData, AgeRangeOptMale = AgeRangeOptMale, AgeRangeOptFemale = AgeRangeOptFemale, BegAgeCompMale = BegAgeCompMale, BegAgeCompFemale = BegAgeCompFemale, ShowPlot = T) ## ---------- If the completion is not satisfying, we modify the values ## ---------- AgeRangeOpt and BegAgeComp, and we repeat the previous script ## ---------- CompletionA() ## ---------- If the completion is satisfying, we execute FinalMethod1 <- CompletionB(ModCompletion = CompletionMethod1, OutputMethod = OutputMethod1, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Validate method 1 by the 3rd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 3rd level criterions ValidationLevel3Method1 <- ValidationLevel3(FinalMethod = FinalMethod1, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Coef Varition, Conf int. and rel. disp. of fitted per. life exp. ## ## ------------------------------------------------------------------------ ## ## ---------- Compute the coefficient of variation, confidence intervals and ## ---------- relative dispersion of the fitted perdiodic life expectancies DispersionMethod1 <- Dispersion(FinalMethod = FinalMethod1, MyData = MyData, Plot = T,NbSim = 10) ## ######################################################################## ## ## METHOD 2 ############################################################### ## ## ######################################################################## ## ## ------------------------------------------------------------------------ ## ## Execute method 2 ## ## ------------------------------------------------------------------------ ## OutputMethod2 <- Method2(MyData = MyData, AgeRange = AgeRange, Plot = T) ## ------------------------------------------------------------------------ ## ## Validate method 2 by the 1st level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 1st level citeria. ValidationLevel1Method2 <- ValidationLevel1(OutputMethod = OutputMethod2, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Plot = T, Excel = T) ## ---------- If the criterions corresponding to the 1st level are not ## ---------- satisfied, we turn to the method 3, and it is useless to ## ---------- pursue the completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the criterions corresponding to the 2nd level. ## ---------- We can also turn to method 4 to improve the fit at a cost ## ---------- of a somewhat greeter complexity. ## ------------------------------------------------------------------------ ## ## Validate method 2 by the 2nd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 2nd level criterions ValidationLevel2Method2 <- ValidationLevel2(OutputMethod = OutputMethod2, AgeCrit = AgeCrit, ValCrit = 0.05, MyData = MyData, Excel = T) ## ---------- If the criterions corresponding to the 2nd level are not satisfied ## ---------- we turn to the method 3 and it is useless to pursue the ## ---------- completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the completion of the table and the criterions corresponding to ## ---------- the 3rd level. ## ------------------------------------------------------------------------ ## ## Completion Method 2 ## ## ------------------------------------------------------------------------ ## ## ---------- We check if the completion is smoothed with graphical ## ---------- diqgnostics. CompletionMethod2 <- CompletionA(OutputMethod = OutputMethod2, MyData = MyData, AgeRangeOptMale = AgeRangeOptMale, AgeRangeOptFemale = AgeRangeOptFemale, BegAgeCompMale = BegAgeCompMale, BegAgeCompFemale = BegAgeCompFemale, ShowPlot = T) ## ---------- If the completion is not satisfying, we modify the values ## ---------- AgeRangeOpt and BegAgeComp, and we repeat the previous script ## ---------- CompletionA() ## ---------- If the completion is satisfying, we execute FinalMethod2 <- CompletionB(ModCompletion = CompletionMethod2, OutputMethod = OutputMethod2, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Validate method 2 by the 3rd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 3rd level criterions ValidationLevel3Method2 <- ValidationLevel3(FinalMethod = FinalMethod2, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Coef Varition, Conf int. and rel. disp. of fitted per. life exp. ## ## ------------------------------------------------------------------------ ## ## ---------- Compute the coefficient of variation, confidence intervals and ## ---------- relative dispersion of the fitted perdiodic life expectancies DispersionMethod2 <- Dispersion(FinalMethod = FinalMethod2, MyData = MyData, Plot = T, NbSim = 10) ## ######################################################################## ## ## METHOD 3 ############################################################### ## ## ######################################################################## ## ## ------------------------------------------------------------------------ ## ## Execute method 3 ## ## ------------------------------------------------------------------------ ## OutputMethod3 <- Method3(MyData = MyData, AgeRange = AgeRange, Plot = T) ## ------------------------------------------------------------------------ ## ## Validate method 3 by the 1st level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 1st level citeria. ValidationLevel1Method3 <- ValidationLevel1(OutputMethod = OutputMethod3, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Plot = T, Excel = T) ## ---------- If the criterions corresponding to the 1st level are not ## ---------- satisfied, we turn to the method 4, and it is useless to ## ---------- pursue the completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the criterions corresponding to the 2nd level. ## ------------------------------------------------------------------------ ## ## Validate method 3 by the 2nd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 2nd level criterions ValidationLevel2Method3 <- ValidationLevel2(OutputMethod = OutputMethod3, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Excel = T) ## ---------- If the criterions corresponding to the 2nd level are not satisfied ## ---------- we turn to the method 4 and it is useless to pursue the ## ---------- completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the completion of the table and the criterions corresponding to ## ---------- the 3rd level. ## ------------------------------------------------------------------------ ## ## Completion Method 3 ## ## ------------------------------------------------------------------------ ## ## ---------- We check if the completion is smoothed with graphical ## ---------- diqgnostics. CompletionMethod3 <- CompletionA(OutputMethod = OutputMethod3, MyData = MyData, AgeRangeOptMale = AgeRangeOptMale, AgeRangeOptFemale = AgeRangeOptFemale, BegAgeCompMale = BegAgeCompMale, BegAgeCompFemale = BegAgeCompFemale, ShowPlot = T) ## ---------- If the completion is not satisfying, we modify the values ## ---------- AgeRangeOpt and BegAgeComp, and we repeat the previous script ## ---------- CompletionA() ## ---------- If the completion is satisfying, we execute FinalMethod3 <- CompletionB(ModCompletion = CompletionMethod3, OutputMethod = OutputMethod3, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Validate method 3 by the 3rd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 3rd level criterions ValidationLevel3Method3 <- ValidationLevel3(FinalMethod = FinalMethod3, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Coef Varition, Conf int. and rel. disp. of fitted per. life exp. ## ## ------------------------------------------------------------------------ ## ## ---------- Compute the coefficient of variation, confidence intervals and ## ---------- relative dispersion of the fitted perdiodic life expectancies DispersionMethod3 <- Dispersion(FinalMethod = FinalMethod3, MyData = MyData, Plot = T, NbSim = 10) ## ######################################################################## ## ## METHOD 4 ############################################################### ## ## ######################################################################## ## ## ------------------------------------------------------------------------ ## ## Execute method 4 ## ## ------------------------------------------------------------------------ ## ## ---------- Execute method 4 first part. OutputMethod4PartOne <- Method4A(MyData = MyData, AgeRange = AgeRange, AgeCrit = AgeCrit, ShowPlot = T) ## ---------- Select the optimal smoothing parameters. ## ---------- Execute method 4 second part. OutputMethod4 <- Method4B(PartOne, MyData = MyData, OptMale = c(1, 16), OptFemale = c(1, 14), Plot = T) ## ------------------------------------------------------------------------ ## ## Validate method 4 by the 1st level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 1st level citeria. ValidationLevel1Method4 <- ValidationLevel1(OutputMethod = OutputMethod4, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Plot = T, Excel = T) ## ---------- If the criterions corresponding to the 1st level are not ## ---------- satisfied, we turn to the method 4, and it is useless to ## ---------- pursue the completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the criterions corresponding to the 2nd level. ## ------------------------------------------------------------------------ ## ## Validate method 4 by the 2nd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 2nd level criterions ValidationLevel2Method4 <- ValidationLevel2(OutputMethod = OutputMethod4, MyData = MyData, AgeCrit = AgeCrit, ValCrit = 0.05, Excel = T) ## ---------- If the criterions corresponding to the 2nd level are not satisfied ## ---------- we turn to the method 4 and it is useless to pursue the ## ---------- completion of the table and the validation. ## ---------- If the criterions are satisfied, we continue the validation with ## ---------- the completion of the table and the criterions corresponding to ## ---------- the 3rd level. ## ------------------------------------------------------------------------ ## ## Completion Method 4 ## ## ------------------------------------------------------------------------ ## ## ---------- We check if the completion is smoothed with graphical ## ---------- diqgnostics. CompletionMethod4 <- CompletionA(OutputMethod = OutputMethod4, MyData = MyData, AgeRangeOptMale = AgeRangeOptMale, AgeRangeOptFemale = AgeRangeOptFemale, BegAgeCompMale = BegAgeCompMale, BegAgeCompFemale = BegAgeCompFemale, ShowPlot = T) ## ---------- If the completion is not satisfying, we modify the values ## ---------- AgeRangeOpt and BegAgeComp, and we repeat the previous script ## ---------- CompletionA() ## ---------- If the completion is satisfying, we execute FinalMethod4 <- CompletionB(ModCompletion = CompletionMethod4, OutputMethod = OutputMethod4, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Validate method 4 by the 3rd level citeria ## ## ------------------------------------------------------------------------ ## ## ---------- Execute 3rd level criterions ValidationLevel3Method4 <- ValidationLevel3(FinalMethod = FinalMethod4, MyData = MyData, Plot = T, Excel = T) ## ------------------------------------------------------------------------ ## ## Coef Varition, Conf int. and rel. disp. of fitted per. life exp. ## ## ------------------------------------------------------------------------ ## ## ---------- Set the number of simulations ## ---------- Compute the coefficient of variation, confidence intervals and ## ---------- relative dispersion of the fitted perdiodic life expectancies DispersionMethod4 <- Dispersion(FinalMethod = FinalMethod4, MyData = MyData, Plot = T, NbSim = 10) ## ######################################################################## ## ## COMPARISON OF THE METHODS ############################################## ## ## ######################################################################## ## ## ---------- Once we have fitted the data with a number of methods, we can ## ---------- compare them. In the following, we compare the fitted ## ---------- probabilities of death in original and log scale, the ## ---------- residuals, the fitted deaths as well as the coherence of the ## ---------- extrapolated mortality trends ## ---------- You can change the color vector for comparison, color need to ## ---------- be in html format ## ---------- Store the output into a list ListOutputs <- list(OutputMethod1, OutputMethod2, OutputMethod3, OutputMethod4) ListValidationLevel1 <- list(ValidationLevel1Method1, ValidationLevel1Method2, ValidationLevel1Method3, ValidationLevel1Method4) ListValidationLevel2 <- list(ValidationLevel2Method1, ValidationLevel2Method2, ValidationLevel2Method3, ValidationLevel2Method4) ListValidationLevel3 <- list(ValidationLevel3Method1, ValidationLevel3Method2, ValidationLevel3Method3, ValidationLevel3Method4) ComparisonsMethodsLevels123 <- ComparisonMethods(ListOutputs, ListValidationLevel1, ListValidationLevel2, ListValidationLevel3, MyData = MyData, Plot = T, AgeCrit = AgeCrit) ## End(Not run) ```

ELT-package