Evidence Analysis with respect to the Target Node (5.0.2)

Analysis of the pieces of evidence (here, the 2 pieces of evidence described in the Analysis Context).

Their joint probability returned the BBN is compared to the one given by the straw model (the fully unconnected network).

Conflict

GC(E) = \log_2 \frac{\prod_{i=1}^n P(e_i \mid e_{i-1}, \ldots, e_1)}{\prod_{i=1}^n P(e_i)}

The reference evidence is the one set to the Target Node.

Evidence Consistency Analysis

The Evidence Consistency Analysis returns the Local Consistency (LC) and the Bayes Factor (BF) for every state of the unobserved variables, given the set of evidence.

BF(E, h) = \log_2 \frac{P(h|E)}{P(h)}

LC(E, h) = \log_2 \frac{\prod_{i=1}^n P(e_i | e_{i-1}, \ldots, e_1) P(h | E)}{\prod_{i=1}^n P(e_i) P(h)}

LC(E, h) = GC(E) + BF(E, h)