Denny's model mortality for lx and its fitting
Implementing Denny's model mortality function of lx and fitting the model to actual lx of given lifetable.
Denny(a, b, c, t) fitDenny(initialpar=rep(0.1, 3), data, mode=3, Method="Nelder-Mead", ...)
a |
The parameter a of the Denny model, l(t)=1/(1+a*(t/(105-t))^3+b*sqrt(exp(t/(105-t))-1)+c*(1-exp(-2*t))). |
b |
The parameter b of the Denny model, l(t)=1/(1+a*(t/(105-t))^3+b*sqrt(exp(t/(105-t))-1)+c*(1-exp(-2*t))). |
c |
The parameter c of the Denny model, l(t)=1/(1+a*(t/(105-t))^3+b*sqrt(exp(t/(105-t))-1)+c*(1-exp(-2*t))). |
t |
Age (vector OK) in years. The t must be less than 105, otherwise the value by Denny() become 0. |
initialpar |
Initial value for the parameters to be estimated. If not given, rep(0.1, 3) is used. |
data |
Actual vector of qx in the lifetable to be used to obtain the best-fit parameters of the Denny's model. If the ages for qx are equal or elder than 105 years old, those will be ignored in fitting. |
mode |
Which of lifetable functions should be used to calculate the RMSE: 1 qx, 2 dx, otherwise lx. Default is 3. |
Method |
The method to be used in optim() function. Default is "Nelder-Mead". |
... |
Other options to be passed to optim(). |
Denny() returns model lx for the same length with t. fitDenny() returns the numeric vector of fitted parameters a, b, and c, RMSE for those values, and the flag of convergence.
Minato Nakazawa minato-nakazawa@umin.net http://minato.sip21c.org/
Denny C (1997) A model of the probability of survival from birth. Mathematical and Computer Modelling, 26: 69-78. doi: 10.1016/S0895-7177(97)00170-2
res <- fitDenny(,qxtolx(Jlife$qx2005M))
FLAG <- res[5]
while (FLAG>0) {
res <- fitDenny(res[1:3], qxtolx(Jlife$qx2005M))
FLAG <- res[5]
}
print(res)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.