public double ClassifyInstanceProba(PmlInstance instance)
{
Build();
var confidences = Impl.distributionForInstance(instance.Impl);
if (confidences.Length != 2)
{
throw new NotSupportedException("ClassifyProba only supports binary classifiers");
}
return(confidences[0] > confidences[1] ?
1 - confidences[0] :
confidences[1]);
}
public void test_creation()
{
ExtendableObj <TestingRow>[] rows = new[] {
new TestingRow {
CLASS = 1.0, ATT_1 = "1"
},
new TestingRow {
CLASS = 2.0, ATT_1 = "2"
}
}.Select(ExtendableObj.Create).ToArray();
rows[0].AddNumerical("additional1", 3);
rows[1].AddNumerical("additional1", 4);
rows[0].AddBinary("additional2", true);
rows[1].AddBinary("additional2", false);
rows[0].AddNumerical("additional3", 5);
rows[1].AddNumerical("additional3", 6);
Runtime rt = Runtime.LoadFromRows(0, rows);
PmlInstance r1 = rt.ElementAt(0);
PmlInstance r2 = rt.ElementAt(1);
Assert.AreEqual(1, r1.Value(0));
Assert.AreEqual(2, r2.Value(0));
Assert.AreEqual(0, r1.Value(1));
Assert.AreEqual(1, r2.Value(1));
Assert.AreEqual(3, r1.Value(2)); // additional1
Assert.AreEqual(4, r2.Value(2));
Assert.AreEqual(1, r1.Value(3)); // additional2
Assert.AreEqual(0, r2.Value(3));
Assert.AreEqual(5, r1.Value(4)); // additional3
Assert.AreEqual(6, r2.Value(4));
}
public double ClassifyInstance(PmlInstance instance)
{
Build();
return(Impl.classifyInstance(instance.Impl));
}
public int ClusterInstance(PmlInstance instance)
{
return(Impl.clusterInstance(instance.Impl));
}
