I'm trying to run a Monte Carlo to determine a set of given weights.
I have 5 weights (w1 to w5) that add up to 100%. Many people have different opinions on what these weights should be. We have collected 40-50 samples of what people view the weights as being. For example:
Person 1: .40 | .30 | .10 | .10 | .10
Person 2: .30 | .20 | .20 | .10 | .20
Person 3: .40 | .25 | .15 | .15 | .15
etc.
I want to run a Monte Carlo by using the data points we've collected and creating a distribution for each weight. Then I will randomly draw a from each distribution over many iterations.
Is this the correct approach?
Additionally, what distribution should I use to model these weights? Two constraints I see:
- no weight can be less than 0
- all weights must add up to 1.0
Any advice or assistance is much appreciated.
