Number of Samples
Specifies the number of samples for the discrete tree simulation. "Initial samples" is the number of samples to draw at the beginning of the tree. If there are no downstream decisions in your model, the total number of samples equals the initial samples. "Minimum at decisions" is the minimum number of samples to draw at downstream decisions. If DPL reaches a downstream decision with fewer than the specified minimum at decisions, then DPL draws additional samples to meet the minimum. Therefore, if there are downstream decisions in your model, the total number of samples may exceed the initial samples.
Multiple Samples
Controls the way the simulation allocates samples at the beginning of the decision tree.
"Pure Monte Carlo simulation" samples using a traditional Monte Carlo simulation technique, i.e., it draws samples completely at random, called perfect sampling.
"Modified Monte Carlo simulation" improves the accuracy of the simulation by using a stratified sampling simulation technique.
"Distributed Sampling" provides the most accurate simulation results by using a combination of perfect sampling and stratified sampling techniques. This is the default sampling method for Discrete Tree Simulation.
Single Samples
Controls the way the simulation handles samples towards the end of the decision tree.
Monte Carlo allocates single samples to randomly selected branches of the decision tree. This is the default method.
Don't Gamble creates a single "average" state for each chance node and allocates single samples to them. This method may provide more accurate results but may also increase run-times.
Versions: DPL Professional, DPL Enterprise, DPL Portfolio
See Also