public static void Main(params string[] args)
{
Parameter param = new Parameter();
// default values
param.svm_type = Parameter.C_SVC;
param.kernel_type = Parameter.RBF;
param.degree = 3;
param.gamma = 0; // 1/num_features
param.coef0 = 0;
param.nu = 0.5;
param.cache_size = 100;
param.C = 1;
param.eps = 1e-3;
param.p = 0.1;
param.shrinking = 1;
param.probability = 1;
param.nr_weight = 0;
param.weight_label = new int[0];
param.weight = new double[0];
Problem problem = SVMFactory.LoadSVMFile("heart_scale.data", param);
Model model = SupportVectorMachine.Train(problem, param);
int errors = 0;
for (int i = 0; i < problem.y.Length; ++i)
{
double[] prob = new double[3];
var newPred = SupportVectorMachine.PredictProbability(model, problem.x[i], prob);
var prediction = SupportVectorMachine.Predict(model, problem.x[i]);
if (prediction != problem.y[i])
{
++errors;
System.Console.ForegroundColor = System.ConsoleColor.Red;
}
else
{
System.Console.ForegroundColor = System.ConsoleColor.White;
}
System.Console.WriteLine("Expected: {0} || Returned: {1}", problem.y[i], prediction);
System.Console.WriteLine("1: {0}% || 2: {1}\n", prob[0] * 100D, prob[1] * 100D);
}
double perc = ((double)errors / problem.y.Length) * 100D;
System.Console.WriteLine("Errors {0} at {1}%", errors, perc);
System.Console.ReadLine();
}
