
More Complex Cases: Hybrid Bayesian Networks 75
PR ∼ Mu (1, p = (0.7, 0.2, 0.1))
CL ∼ Beta (3, 1)
G1 | PR = p, CL = c ∼ P ois (c × g (p)))
G2 | G1 = g
1
∼ Pois (10g
1
)
TR | G1 = g
1
∼ Ber
logit
−1
g
1
− 5
2.5
LO | G2 = g
2
, TR = t ∼ ncχ
2
1,
g
2
×
1 −
2t
3
2
!
Table 3.2
Probability distributions proposed for the DAG shown in Figure 3.4. g is
a known function giving the potential of G1 for a given class of the last
crop: g(1) = 1, g(2) = 3, g(3) = 10. Ber denotes a Bernoulli distribution, P ois
denotes a Poisson distribution and ncχ
2
denotes a non-central Chi-square
distribution (see Appendix B for their definitions and fundamental prope rties).
parameters of the distributions ...