LinearLink: Fit the Linear-Link Model

In mpascariu/MortalityEstimate: Indirect Estimation Methods for Measurement of Demographic Indicators

Description

Calibrate the Linear-Link mortality model introduced in \insertCitepascariu2020;textualMortalityEstimate.

Usage

LinearLink(
  x,
  mx,
  y,
  country = "...",
  theta = 0,
  use.smooth = TRUE,
  method = "LSE"
)

Arguments

x	Numerical vector containing ages corresponding to the input data (mx).
mx	Death rates matrix with age as row and time as column.
y	Vector of years corresponding to the mx matrix.
country	Optional. The name of the country that the data corresponds to. The name is adopted in the output tables.
theta	Age to be fitted.
use.smooth	Logical variable indicating whether the spline smoothing is applied or not to the estimated coefficients (bx and vx). The smoothing can be applied in order to avoid jumps in the mortality rates from one age to another. This using splines. One degree of freedom is allocated for every 5 year of age.
method	Optimizing method. Least squared approach LSE or Poisson likelihood estimation MLE based on the approach described in \insertCitebrouhns2002;textualMortalityEstimate. Default: LSE.

Value

A LinearLink object containing:

input	List with input objects provided in the function
coefficients	Estimated coefficient
fitted.values	Fitted values
residuals	Estimated deviance residuals
fitted.life.tables	Life tables constructed using the fitted values
df_splines	Degrees of freedom used in spline smoothing procedure
model_info	Description of the model
process_date	Data and time stamp

References

\insertAllCited

Pascariu MD (2018). PhD Thesis: Modelling and Forecasting Mortality. University of Southern Denmark, 53-70. URL: https://github.com/mpascariu/PhD-Thesis/blob/master/Thesis.pdf.

Examples

# Select the 1965 - 1990 time interval and fit the Linear-Link model
ages  <- 0:100 # available ages in our datasets
years <- 1965:1990 # available years
sex   <- 'female'
SWEmx <- HMD4mx$SWE[paste(ages), paste(years)]

# Fit the Linear-Link using the least square approach (LSE).
M <- LinearLink(x  = ages,
                mx = SWEmx,
                y  = years,
                country = 'SWEDEN',
                theta   = 0,
                method  = 'LSE')
M
summary(M)
coef(M)
ls(M)


# Derive a mortality curve (life table) from a value of
# life expectancy at birth in 2014, say 84.05
e0 <- 84.05
LT1 <- LinearLinkLT(M, ex = e0)
LT2 <- LinearLinkLT(M, ex = e0, use.vx.rotation = TRUE) 

mpascariu/MortalityEstimate documentation built on May 11, 2021, 6:33 p.m.