Refining Influence Diagram For Stock Portfolio Selection

In this paper, we propose use of the influence diagram for stock portfolio selection. We use an algorithm that applies the mutual information as the metric to guide the refinement of the influence diagram. We applied the algorithm to the conceptual refinement of the influence diagram. We tested our algorithm to a specific domain ƒV portfolio selection; the result is impressive compared to the S&P500. Experiments comparing our model with the other machine learning techniques are also provided.