public static void SaveKnn(KNearestNeighbors knn)
{
// After we have created and learned our model, let's say we would
// like to save it to disk. For this, we can import the Accord.IO
// namespace at the top of our source file namespace, and then use
// Serializer's extension method Save:
// Save to a file called "knn.bin" in the basePath directory:
knn.Save(Path.Combine(basePath, "knn.bin"));
}
/// <summary>
/// Classify our data using k-nearest neighbors classifer and save the model.
/// </summary>
/// <param name="train_data">Frame objects that we will use to train classifers.</param>
/// <param name="test_data">Frame objects that we will use to test classifers.</param>
/// <param name="train_label">Labels of the train data.</param>
/// <param name="test_label">Labels of the test data.</param>
/// <param name="Classifier_Path">Path where we want to save the classifer on the disk.</param>
/// <param name="Classifier_Name">Name of the classifer we wnat to save.</param>
/// <returns></returns>
public void Knn(double[][] train_data, double[][] test_data, int[] train_label, int[] test_label, String Classifier_Path, String Classifier_Name)
{
KNearestNeighbors knn = new KNearestNeighbors(k: 5);
knn.Learn(train_data, train_label);
int answer = knn.Decide(new double[] { 117.07004523277283, 119.9104585647583 });
var cm = GeneralConfusionMatrix.Estimate(knn, test_data, test_label);
double error = cm.Error;
Console.WriteLine(error);
knn.Save(Path.Combine(Classifier_Path, Classifier_Name));
}
public void SaveLearnedModel()
{
KnnModel.Save(Path.Combine(Constants.BasePath, _experimentName + ".bin"));
}
private void saveModel()
{
classifier.Save(modelPath);
}
public void learn_string()
{
string basePath = NUnit.Framework.TestContext.CurrentContext.TestDirectory;
#region doc_learn_text
// The k-Nearest Neighbors algorithm can be used with
// any kind of data. In this example, we will see how
// it can be used to compare, for example, Strings.
string[] inputs =
{
"Car", // class 0
"Bar", // class 0
"Jar", // class 0
"Charm", // class 1
"Chair" // class 1
};
int[] outputs =
{
0, 0, 0, // First three are from class 0
1, 1, // And next two are from class 1
};
// Now we will create the K-Nearest Neighbors algorithm. For this
// example, we will be choosing k = 1. This means that, for a given
// instance, only its nearest neighbor will be used to cast a new
// decision.
// In order to compare strings, we will be using Levenshtein's string distance
var knn = new KNearestNeighbors <string>(k: 1, distance: new Levenshtein());
// We learn the algorithm:
knn.Learn(inputs, outputs);
// After the algorithm has been created, we can use it:
int answer = knn.Decide("Chars"); // answer should be 1.
// Let's say we would like to compute the error matrix for the classifier:
var cm = ConfusionMatrix.Estimate(knn, inputs, outputs);
// We can use it to estimate measures such as
double error = cm.Error; // should be 0
double acc = cm.Accuracy; // should be 1
double kappa = cm.Kappa; // should be 1
#endregion
Assert.AreEqual(1, answer);
Assert.AreEqual(0, error);
Assert.AreEqual(1, acc);
Assert.AreEqual(1, kappa);
#if !NO_BINARY_SERIALIZATION
knn.Save(Path.Combine(basePath, "string_knn.bin"));
var loaded_knn = Serializer.Load <KNearestNeighbors <string> >(Path.Combine(basePath, "string_knn.bin"));
Assert.AreEqual(1, loaded_knn.Decide("Chars"));
cm = ConfusionMatrix.Estimate(loaded_knn, inputs, outputs);
Assert.AreEqual(0, cm.Error);
Assert.AreEqual(1, cm.Accuracy);
Assert.AreEqual(1, cm.Kappa);
Assert.AreEqual(knn.ClassCount, loaded_knn.ClassCount);
Assert.AreEqual(knn.Distance, loaded_knn.Distance);
Assert.AreEqual(knn.K, loaded_knn.K);
Assert.AreEqual(knn.NumberOfClasses, loaded_knn.NumberOfClasses);
Assert.AreEqual(knn.NumberOfInputs, loaded_knn.NumberOfInputs);
Assert.AreEqual(knn.NumberOfOutputs, loaded_knn.NumberOfOutputs);
Assert.AreEqual(knn.Outputs, loaded_knn.Outputs);
Assert.AreEqual(knn.Token, loaded_knn.Token);
#endif
}
public void learn_test1()
{
string basePath = NUnit.Framework.TestContext.CurrentContext.TestDirectory;
#region doc_learn
// Create some sample learning data. In this data,
// the first two instances belong to a class, the
// four next belong to another class and the last
// three to yet another.
double[][] inputs =
{
// The first two are from class 0
new double[] { -5, -2, -1 },
new double[] { -5, -5, -6 },
// The next four are from class 1
new double[] { 2, 1, 1 },
new double[] { 1, 1, 2 },
new double[] { 1, 2, 2 },
new double[] { 3, 1, 2 },
// The last three are from class 2
new double[] { 11, 5, 4 },
new double[] { 15, 5, 6 },
new double[] { 10, 5, 6 },
};
int[] outputs =
{
0, 0, // First two from class 0
1, 1, 1, 1, // Next four from class 1
2, 2, 2 // Last three from class 2
};
// Now we will create the K-Nearest Neighbors algorithm. For this
// example, we will be choosing k = 4. This means that, for a given
// instance, its nearest 4 neighbors will be used to cast a decision.
var knn = new KNearestNeighbors(k: 4);
// We learn the algorithm:
knn.Learn(inputs, outputs);
// After the algorithm has been created, we can classify a new instance:
int answer = knn.Decide(new double[] { 11, 5, 4 }); // answer will be 2.
// Let's say we would like to compute the error matrix for the classifier:
var cm = GeneralConfusionMatrix.Estimate(knn, inputs, outputs);
// We can use it to estimate measures such as
double error = cm.Error; // should be
double acc = cm.Accuracy; // should be
double kappa = cm.Kappa; // should be
#endregion
Assert.AreEqual(2, answer);
Assert.AreEqual(0, error);
Assert.AreEqual(1, acc);
Assert.AreEqual(1, kappa);
#if !NO_BINARY_SERIALIZATION
#region doc_serialization
// After we have created and learned our model, let's say we would
// like to save it to disk. For this, we can import the Accord.IO
// namespace at the top of our source file namespace, and then use
// Serializer's extension method Save:
// Save to a file called "knn.bin" in the basePath directory:
knn.Save(Path.Combine(basePath, "knn.bin"));
// To load it back from the disk, we might need to use the Serializer class directly:
var loaded_knn = Serializer.Load <KNearestNeighbors>(Path.Combine(basePath, "knn.bin"));
// At this point, knn and loaded_knn should be
// two different instances of identical objects.
#endregion
// Make sure the loaded classifier is still working
Assert.AreEqual(2, loaded_knn.Decide(new double[] { 11, 5, 4 }));
cm = GeneralConfusionMatrix.Estimate(loaded_knn, inputs, outputs);
Assert.AreEqual(0, cm.Error);
Assert.AreEqual(1, cm.Accuracy);
Assert.AreEqual(1, cm.Kappa);
Assert.AreEqual(knn.ClassCount, loaded_knn.ClassCount);
Assert.AreEqual(knn.Distance, loaded_knn.Distance);
Assert.AreEqual(knn.K, loaded_knn.K);
Assert.AreEqual(knn.NumberOfClasses, loaded_knn.NumberOfClasses);
Assert.AreEqual(knn.NumberOfInputs, loaded_knn.NumberOfInputs);
Assert.AreEqual(knn.NumberOfOutputs, loaded_knn.NumberOfOutputs);
Assert.AreEqual(knn.Outputs, loaded_knn.Outputs);
Assert.AreEqual(knn.Token, loaded_knn.Token);
#endif
}