Gamma.dist

[wickblue]

Hi,

I keep running a gamma dist with values:

x= 1500
alpha = 320
beta = 0.222
cumulative = TRUE

and I get 1 every single time as the answer.

this cant be possible, as there is not a 100% chance of x=1500 with a k=alpha=320 and a theta=(1/beta)=4.5.

help please.

 

[ChrisM]

I am not a statistician, so don't get all technical on me, but I do use gamma.dist in my job.

You will always get the same answer when you give the same input values. Think of a gamma curve, and imagine the x value you specified is a position along the x axis. The alpha/beta dictate the shape of the curve, and the x value dictates how far along the curve you are, so you always will get the same answer.

So by the time x hits 1500, your function is telling you that 100% of the population is covered. If you want to get a number less than 100%, you need to lower your x value. But the other issue is your alpha, 320 seems crazy high, you are going to get a really skewed curve. Given your current alpha/beta, any x value greater than 82 will give you a 100% answer.

If you use an x less than 82, take 71 for example, THEN you can start to say "at 71, there is a 50% chance of whatever, at 75 there is an 84% chance, etc". But your x is so large, and your alpha so large, you have a really weird scenario here.

 

[gsistek]

Also, since you have culmulative set to true, you are not calculating the probability of x=1500, you are calculating the probability of x < =1500, which is why you are getting 1. Though, even calculating Pr(x=1500) with those parameters yields 1.23E(-2374).