static void Main(string[] args)
{
SVMProblem testSet = SVMProblemHelper.Load(@"Dataset\wine.txt"); // Same as the training set
SVMModel model = SVM.LoadModel(@"Model\wine_model.txt");
Console.WriteLine("Feature count in one instance: " + model.SV[0].Length + "\n\n");
// Test 1: Predict instances with SVMProblem's Predict extension method.
sw.Start();
double[] target = testSet.Predict(model);
sw.Stop();
double elapsedTimeInTest1 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 1: \nPredict instances with SVMProblem's Predict extension method.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest1 + " ms\n");
// Test 2: Predict instances with RapidPreditor class which is an explicit implementation of the method used in Test 1.
using (RapidPredictor predictor = new RapidPredictor(model)) // It needs to be Disposed
{
sw.Start();
target = new double[testSet.Length];
for (int i = 0; i < testSet.Length; i++)
{
target[i] = predictor.Predict(testSet.X[i]);
}
sw.Stop();
}
double elapsedTimeInTest2 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 2: \nPredict instances with RapidPreditor class which is an explicit implementation of the method used in Test 1.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest2 + " ms\n");
// Test 3: Predict instances with standard SVM.Predict method or SVMNode[]'s predict extension method.
sw.Start();
target = new double[testSet.Length];
for (int i = 0; i < testSet.Length; i++)
{
target[i] = SVM.Predict(model, testSet.X[i]);
}
sw.Stop();
double elapsedTimeInTest3 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 3: \nPredict instances with standard SVM.Predict method or SVMNode[]'s Predict extension method.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest3 + " ms\n");
// Print the results
Console.WriteLine("\nExplanation:\n");
Console.WriteLine(
"In standard SVM.Predict method, the SVMModel object is allocated and deallocated every time when the method called. " +
"Also the SVMNode[]'s Predict extension methods directly calls the SVM.Predict. " +
"However, the model is allocated once and is used to predict whole instances with its pointer in SVMProblem's " +
"Predict extension method as implemented in the RapidPredictor class. You can take or modify this class in order " +
"to use in your applications, if you have performance considerations. " +
"I am not suggesting that SVMProblem's Predict extension method is used in real-time, because the model is allocated" +
"in every method call.");
Console.WriteLine("\n\nPress any key to quit...");
Console.ReadLine();
}
static void Main(string[] args)
{
SVMProblem testSet = SVMProblemHelper.Load(@"Dataset\wine.txt"); // Same as the training set
SVMModel model = SVM.LoadModel(@"Model\wine_model.txt");
Console.WriteLine("Feature count in one instance: " + model.SV[0].Length +"\n\n");
// Test 1: Predict instances with SVMProblem's Predict extension method.
sw.Start();
double[] target = testSet.Predict(model);
sw.Stop();
double elapsedTimeInTest1 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 1: \nPredict instances with SVMProblem's Predict extension method.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest1 + " ms\n");
// Test 2: Predict instances with RapidPreditor class which is an explicit implementation of the method used in Test 1.
using (RapidPredictor predictor = new RapidPredictor(model)) // It needs to be Disposed
{
sw.Start();
target = new double[testSet.Length];
for (int i = 0; i < testSet.Length; i++)
target[i] = predictor.Predict(testSet.X[i]);
sw.Stop();
}
double elapsedTimeInTest2 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 2: \nPredict instances with RapidPreditor class which is an explicit implementation of the method used in Test 1.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest2 + " ms\n");
// Test 3: Predict instances with standard SVM.Predict method or SVMNode[]'s predict extension method.
sw.Start();
target = new double[testSet.Length];
for (int i = 0; i < testSet.Length; i++)
target[i] = SVM.Predict(model, testSet.X[i]);
sw.Stop();
double elapsedTimeInTest3 = (double)sw.ElapsedMilliseconds / (double)testSet.Length;
Console.WriteLine("> Test 3: \nPredict instances with standard SVM.Predict method or SVMNode[]'s Predict extension method.\n");
Console.WriteLine("\tAverage elapsed time of one prediction: " + elapsedTimeInTest3 + " ms\n");
// Print the results
Console.WriteLine("\nExplanation:\n");
Console.WriteLine(
"In standard SVM.Predict method, the SVMModel object is allocated and deallocated every time when the method called. " +
"Also the SVMNode[]'s Predict extension methods directly calls the SVM.Predict. " +
"However, the model is allocated once and is used to predict whole instances with its pointer in SVMProblem's " +
"Predict extension method as implemented in the RapidPredictor class. You can take or modify this class in order " +
"to use in your applications, if you have performance considerations. " +
"I am not suggesting that SVMProblem's Predict extension method is used in real-time, because the model is allocated" +
"in every method call.");
Console.WriteLine("\n\nPress any key to quit...");
Console.ReadLine();
}
