Improvement of Adequate Digoxin Dosage: An Application of Machine Learning Approach

Table 3

Parameter settings in WEKA.

Method	Parameters	Value/Range	Best parameter setting
J48	Confidence factor	0.1–0.5	0.25
	Minimum number of instances per leaf	2–50	2
IBk	Number of neighbors	2–10	2
SimpleCART	Minimum number of instances per leaf	2–50	2
RandomForest	Number of trees	5–10	10
	Number of attributes to be used in random selection	2–8	4
Multilayer perceptron	Number of hidden nodes	3–14	7
	Learning rate	0.1–0.6	0.3
	Momentum factor	0–0.9	0.2
	Maximum number of epochs	300–1000	500
AdaBoostM1	Number of iterations	10	10
	Weight threshold for pruning	100	100