The Value Correlation Chart displays the correlation coefficient between the objective function and each value in the model. This chart is useful for models with complex value models to understand how strongly correlated each value is to objective function outcomes.
The correlation coefficient is between negative one and positive one.
The closer to positive one the correlation coefficient is the more strongly correlated the value is to the objective function. I.e., when the objective function is high, the value is more likely to be high and when the objective function is low, the value is more likely to be low.
The closer to negative one the correlation coefficient is the more strongly oppositely correlated the value is to the objective function. I.e., when the objective function is high, the value is more likely to be low and when the objective function is low, the value is more likely to be high.
A correlation coefficient of zero means the value is uncorrelated with the objective function. Correlation coefficients near zero mean the value is only weakly correlated with the objective function.
Value correlations can only be calculated for values that can be evaluated on all paths of the tree. E.g., if a value is conditioned by one or more events that do not occur on certain paths, then it can not be evaluated on those paths and a value correlation is not calculated for it.
You can choose to run a Value Correlation Chart when you run a Decision Analysis or Play Endpoints. When you run a Decision Analysis and select Value Correlations, full tree enumeration is used as the evaluation method.
Versions: DPL Professional, DPL Enterprise, DPL Portfolio
See Also