**Go backward to **Solution to part (d)

**Go up to **4 Variable Elimination Algorithm (Multiply Connected)

**Go forward to **Solution to part (f)

## Solution to part (e)

Suppose, without any observations, we eliminate *F*. What
factors are removed and what factor is created. Give a general rule as
to what variables are joined when a variable is eliminated from a
Bayesian network.
We remove the factors *P(F|D)*, *P(H|E,F)*, *P(I|F,G)* and replace
them with *f*_{1}(D,E,H,I,G).

The general rule is that we join the parents of the node being
removed, the children of the node being removed and the children's
other parents.

Computational
Intelligence online
material, ©David Poole, Alan Mackworth and Randy Goebel, 1999