do.MortalityModels: Fit Multiple Stochastic Mortality Models

In mpascariu/MortalityForecast: Standard Tools to Compare and Evaluate Various Mortality Forecasting Methods

Description

Fit Multiple Stochastic Mortality Models

Usage

do.MortalityModels(data, data.B = NULL, x = NULL, y = NULL,
  data.in = c("qx", "mx", "dx", "lx"), models = c("MRWD"),
  verbose = TRUE, ...)

Arguments

data	A data.frame or a matrix containing mortality data with ages x as row and time y as column.
data.B	A data.frame or a matrix containing mortality data for the benchmark population. This dataset is needed only in the coherent mortality models (e.g. LiLee, OeppenC). Must be the same format as in data;
x	Numerical vector indicating the ages in input data. Optional. Default: NULL.
y	Numerical vector indicating the years in input data. Optional. Default: NULL.
data.in	Specify the type of input data. Various life table indices are accepted: "qx", "mx", "dx", "lx".
models	One or several mortality models to be estimated. The following options are available: "MRW" – The Multivariate Random-Walk (w/o Drift); "MRWD" – The Multivariate Random-Walk with Drift; "LeeCarter" – The Lee-Carter Mortality Model; "LiLee" – The Li-Lee Mortality Model; "HyndmanUllah" – The Hyndman-Ullah Mortality Model; "RenshawHaberman" – The Renshaw-Haberman Mortality Model; "Oeppen" – The Oeppen Mortality Model; "OeppenC" – The Coherent Oeppen Mortality Model; "MEM2" – The Maximum Entropy Mortality Model of order 2; "MEM3" – The Maximum Entropy Mortality Model of order 3; "MEM4" – The Maximum Entropy Mortality Model of order 4; "MEM5" – The Maximum Entropy Mortality Model of order 5; "MEM6" – The Maximum Entropy Mortality Model of order 6; "MEM7" – The Maximum Entropy Mortality Model of order 7;
verbose	A logical value. Set verbose = FALSE to silent the process that take place inside the function and avoid progress messages.
...	Arguments to be passed to or from other methods.

Examples

x  <- 0:100      # Ages
y  <- 2005:2016  # Years
h  <- 16         # forecasting horizon
MM <- c("MRWD", "LeeCarter", "LiLee", "HyndmanUllah", 
        "Oeppen", "OeppenC", "MEM5", "RenshawHaberman") # mortality models
D  <- HMD_male$dx$GBRTENW[paste(x), paste(y)]  # data
B  <- HMD_female$dx$GBRTENW[paste(x), paste(y)]  # benchmark population

# Note: We are fitting various mortality model to E&W males data. Most of them 
# are single population model i.e. the estimates are resulted only from data 
# specific to that population. However, the coherent models like "OeppenC" or 
# "LiLee" require additional data. In these cases, here, female data is used.


M <- do.MortalityModels(data = D, 
                        data.B = B,
                        x = x, 
                        y = y, 
                        data.in = "dx", 
                        models = MM)

P <- do.MortalityForecasts(object = M, 
                           h = h, 
                           level = 95, 
                           jumpchoice = "actual")

oex <- get.Observed(M, data.out = "ex")
fex <- get.Fitted(M, data.out = "ex")
rex <- get.Residuals(M, data.out = "ex")
pex <- get.Forecasts(P, data.out = "ex")

mpascariu/MortalityForecast documentation built on Sept. 28, 2020, 2:40 p.m.