Research Article

Noninvasive Evaluation of Portal Hypertension Using a Supervised Learning Technique

Table 4

The results of all four algorithms test scenarios.

Scenario IScenario IIScenario IIIScenario IV
AlgorithmAUCSTDERRt-testAUCSTDERRt-testAUCSTDERRt-testAUCSTDERRt-test

Bayes Net0.880.04Base0.900.03Base0.900.04Base0.890.03Base
Naive Bayes0.950.0200.950.0300.940.0400.920.020
Logistic0.960.02Base0.940.02Base0.960.02Base0.930.02Base
Multilayer Perceptron0.900.0300.950.0200.950.0200.910.020
SGD0.820.0500.890.03−10.870.04−10.820.03−1
Simple Logistic0.950.0200.950.0200.950.0300.930.020
SMO0.870.0500.860.04−10.830.04−10.750.04−1
Voted Perceptron0.950.0400.910.0300.930.0300.790.030
LazyIBk0.910.03Base0.880.04Base0.930.04Base0.880.03Base
lazy.Kstar0.970.0210.970.0210.940.0300.840.020
Decision Table0.890.03Base0.880.03Base0.880.04Base0.850.04Base
JRip0.820.0400.810.0400.800.0500.800.040
OneR0.790.0400.740.0400.790.05−10.800.040
Part0.840.0400.860.0300.910.0400.840.050
Decision Stump0.820.03Base0.820.04Base0.810.04Base0.820.03Base
Hoeffding Tree0.930.0510.900.0200.910.0200.880.040
J480.860.0500.830.0400.930.0400.780.040
LMT0.950.0210.910.0210.950.0310.930.021
Random Forest0.960.0210.940.0210.910.0210.890.021
Random Tree0.770.0500.800.0400.810.0500.800.040
Rep Tree0.870.0400.880.0400.840.0500.840.040