Re: Time to failure problem

Good morning, Nic...

While I'm thinking about it, I meant to mention the following.
I'm assuming that in your data Age 50 = 0.020% implies
(to me) that at age 50 the rate at which this event occurs
is 0.0002 or 2 in 10,000 per year. This can be understood
as a risk, or a probability. But it can also be viewed as a
rate.

Well, I'm not sure of what you are trying to do. What we have at this
point
is a direct way of calculating the probability (or rate) for the cited
event
by (on or before), say, age 85 (during their 85th year) for a person
who is
currently age 50. (prob = 0.106) If that person actually achieves age
84
without experiencing the cited event and is now entering age 85 then
the
corresponding probability of experiencing such an event during his/her
85th year is only prob = 0.0127.

If the subject is currently age 70 (and has not experienced such an
event)
then the prob of experiencing it by the end of their 85th year is not
the same
as beginning from age 50.

The point is that we have a direct way of calculating these rates or
probabilities. We really don't need a "random number" simulation
(or Monte Carlo simulator) to do this. Just run the calculation from
the intial age and terminate it at age 85 or age 86 or whatever you
wish.

However, in your original post you mentioned that your population has
a mean of 70 and a standard deviation of (about) 7 I don't recall the
precise
value for the std. dev. You were assuming that your population is
normally distributed, I believe. Here's where I'm confused/lost.
Do you want to generate random normal numbers N(70, 7.?) for the
initial age values and then directly calculate the probs for the end
points
to get the distriubtion of those end points? If so, then you need a
random number generator for N(70, 7.?) to generate the initial values.
This, too, could be calculated directly... but here I'd go with a Monte
Carlo simulation. If you need such a generator I can post that in this
thread. OMU

Nicbrez wrote:
OMU, thanks for your help with this. Its starting to make a bit more sense. For the simulation, do i then generate a random number between 0 and 1 to get an event from the cumilitve probability of an event. So for example a if i sample 0.87, this relates to no event whereas 0.001 would be an event aged 54?

This may seem an odd question and im not sure its even valid to ask it, but is there anywhat i can fit a distribution to the cumulitive probability so that i can just sample for the distribution rather than having to see where my random number falls in the transition matrix. (does that make sense!!!)

Cheers

Nic

.

Relevant Pages

  • Re: WAYTTGOPTLTL?
    ... the stuff painted by dead people because it "belongs to another age"? ... The reason for experiencing anything new is precisely because it _doesn't_ ...
    (rec.music.classical.recordings)
  • Re: WAYTTGOPTLTL?
    ... when they get to the stuff painted by dead people because it ... "belongs to another age"? ... The reason for experiencing anything new is precisely because it _doesn't_ ...
    (rec.music.classical.recordings)
  • Re: So You Think You Can Dance
    ... Experiencing the Cocoon phenomenon? ... I don't think you really know the age of the person ... that you are commenting about. ...
    (rec.arts.dance)
  • Re: Forgetting stuff
    ... I have lost gear through the years but never a guitar or other ... instrument. ... I don't think what you are experiencing is normal at any age. ...
    (alt.guitar.bass)
  • Re: formatting a calculated number field on a form
    ... > I only used age at death as an example. ... > and heart attack, age 40 and heart attack, surgery and death, etc. ... >>bit different in that a parameter query limits the recordset before it ... I assumed that you were calculating on ...
    (microsoft.public.access.forms)
Quantcast