[wd40bomber7]

I am trying to figure out how to implement a function which will take a list of possibilities and a list of weights and spit out a random possibility.

So if I had three possibilities: A, B, C
and I assigned them the weights: .01,.8,.19

From a sample of the functions output, I would expect to see mostly A, a few B, and maybe one C

The language I'm using is both obscure and irrelevant. The tools I have to use are basic arrays, (Not an arraylist or a vector or anything like that) and a GetRandomReal() function. Also the primitive types real,int,bool

Anyone have any ideas? If you want to write an example, I don't care what language its in as long as its fairly popular and not incredibly difficult to understand.

This post has been edited by wd40bomber7: 22 May 2012 - 03:31 PM

[macosxnerd101]

If your precision on the probabilities is to the hundreths place, generate randoms from 0-99. So 0 would be associated with A, 1-80 would be for B, and 81-99 would be for C.

[wd40bomber7]

That definitely seems like the best option. I don't know why that didn't occur to me right off the bat.

-Thanks

[macosxnerd101]

Glad I could help!

[turboscrew]

Or if you need some more exotic distribution, generate a table with as many similar entries as the proportional share implies.

Then, take a flat distribution random and use the output if it to indicate (calculate) the table index.

That's good way especially if you don't know the distribution, but you can get samples from the real world.

In case you use macosxnerd101's way, remember not to map the seed to the random. It tends to break the random-function logic (pseudo-random sequence).
Use a separate variable.

This post has been edited by turboscrew: 23 May 2012 - 07:35 AM

[baavgai]

Since your function is GetRandomReal, I'd probably keep it real. Assuming the function returns between 0 and 1, and since your weights are already normalized to equal one...

n = GetRandomReal()
if n<0.01:
   A
elif n<0.81:
   B
else:
   C