Biology Reference
In-Depth Information
+
graphviz.plot(hc(dmarks), main = main)
+}
(b) Interval discretization's performance is comparable with the one of quantile
discretization. Again, as the number of intervals increases, the dependence re-
lationships linking the variables are lost, and fewer and fewer arcs are picked
up by the structure learning algorithm.
> intervals = 2:5
> par(mfrow = c(1, length(intervals)))
> for (int in intervals) {
+
dmarks = discretize(marks, breaks = int,
+
method = "interval")
+
main = paste("dmarks,", int, "intervals")
+
graphviz.plot(hc(dmarks), main = main)
+}
(c) Hartemink's discretization performs better than both interval and quantile dis-
cretization if we start with a suitably high number of breaks (
ibreaks = 50
)
and perform the initial discretization with a quantile transform (
idisch =
"interval"
). Even when each variable is discretized into 4 intervals, a sub-
stantial part of the network structure is still learned correctly.
> intervals = 2:5
> par(mfrow = c(1, length(intervals)))
> for (int in intervals) {
+
dmarks = discretize(marks, breaks = int,
+
method = "hartemink", ibreaks = 50,
+
idisc = "interval")
+
main = paste("dmarks,", int, "intervals")
+
graphviz.plot(hc(dmarks), main = main)
+}
2.4
The ALARM network (
Beinlich et al. 1989
) is a Bayesian network designed
to provide an alarm message system for patients hospitalized in intensive care
units (ICU). Since ALARM is commonly used as a benchmark in literature, a
synthetic data set of
5000
observations generated from this network is available
from bnlearn as
alarm
.
(a) Create a
bn
object for the “true” structure of the network using the model
string provided in its manual page.
(b) Compare the networks learned with different constraint-based algorithms
with the true one, both in terms of structural differences and using either
BIC or BDe.
(c) The overall performance of constraint-based algorithms suggests that the
asymptotic
2
conditional independence tests may not be appropriate for
analyzing
alarm
. Are permutation or shrinkage tests better choices?
(d) How are the above learning strategies affected by changes to
alpha
?
χ
Search WWH ::
Custom Search