[R] graduation

[David Winsemius]

On Sep 23, 2009, at 8:44 AM, David Winsemius wrote:

>
> On Sep 23, 2009, at 4:54 AM, MKHABELA,SN wrote:
>
>> Hi everyone
>>
>> I want help in graduating the attached rates and checking for  
>> goodness of fit and smoothness using R please help.
>>
>> <Mortality rates for males and  
>> females.txt>______________________________________________
>
> You have provided the rates but not the death counts or person-years  
> of exposure that would be needed to create these rates. Such numbers  
> would be useful in doing a graduation that had a statistical basis.  
> At the moment I think you should be looking at the data and deciding  
> whether you can model as a Gompertz function which is the canonical  
> "mortality law":
>
> mrates <- read.table(file="<path>/Mortality rates for males and  
> females.txt", header=T)
> with(mrates, plot(Age, log(Males+.001)))  # added 1/2 the minimum  
> rate to the zero entries
> with(mrates, plot(Age, log(females+.0005)))

My "eyeball minimum" wasn't very good and should have been done thusly:

with(mrates, plot(Age, log(females + 0.5*min(mrates$females) )) )

>
> But I do not think you have given enough information or background  
> to take further steps. The plots are not entirely supportive of a  
> Gompertz model and is "telling" you that you need to think more  
> deeply about what the underlying situation really represents.
>
> Of course, if this is homework, you have not been completely  
> forthright with us and should review your educational  institution's  
> regulations.
>

David Winsemius, MD
Heritage Laboratories
West Hartford, CT