And the set d-separations the dag subset ind
|
559 | |||
|---|---|---|---|---|
| X | W | X | ||
| Y | Z | Y | ||
| (a) | (b) | |||
Example 10.13 Suppose V = {X, Y, Z, W} and
IND = {I({X}, {Z}|{Y, W}), I({Y }, {W}|{X, Z})}.
| 560 | X | ||||
|---|---|---|---|---|---|
| H1 |
|
||||
| Y |
|
W | |||
| X | Y | Z | W |
|---|
Figure 10.15: Given V = {X, Y, Z, W, T} and IND consists of the d-separations in the DAG in (a) restricted to these variables, Algorithm 10.4 will produce the DAG in (b) or one with W → T instead of W ← T.
pattern containing the d-separations. The next example shows a situation in which the output of Algorithm 10.4 is a DAG, but the set of d-separations in the DAG is not a subset of IND.
