Q. I thought the Structural Coefficient was the weighting factor on Complexity? So 0.4 means a less complex model with respect to the data fit?

A. You can consider the Structural Coefficient 𝛼 a "penalty weight." The MDL score is computed as follows: MDL(B,D)=𝛼⋅DL(B)+DL(D|B). So, lowering 𝛼 below 1 increases the complexity of the network.

Please also see the BayesiaLab Library in this regard:
Quote 0 0