public void TestBucketEntriesMultipleInputs()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
List <Object> classification = new List <object>();
classification.Add(4);
classification.Add(34.7);
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
classification[0] = 2;
classification[1] = 35.7;
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
classification[0] = 3;
classification[1] = 36.7;
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
classification[0] = 4;
classification[1] = 37.7;
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
classification[0] = 5;
classification[1] = 38.7;
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
Assert.AreEqual(4, sdr.bucketEntries.Count);
Assert.AreEqual(34.7, sdr.bucketEntries[4][0]);
Assert.AreEqual(35.7, sdr.bucketEntries[2][0]);
Assert.AreEqual(36.7, sdr.bucketEntries[3][0]);
Assert.AreEqual(37.7, sdr.bucketEntries[4][1]);
Assert.AreEqual(38.7, sdr.bucketEntries[5][0]);
}
public void TestObjectCreated()
{
double alpha = 0.001;
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>(alpha);
Assert.IsNotNull(sdr);
}
public void TestGrowingMatrixAccordingToBuckets()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
List <Object> classification = new List <object>();
classification.Add(0);
classification.Add(34.7);
int recordNumber = 0;
sdr.Compute(recordNumber, classification, new int[] { 0, 6, 9, 11 });
Assert.IsTrue(sdr.weightMatrix.Matrix.Count == 1);
classification[0] = 4;
classification[1] = 34.7;
sdr.Compute(1, classification, new int[] { 0, 6, 9, 11 });
Assert.IsTrue(sdr.weightMatrix.Matrix.Count == 5);
classification[0] = 6;
classification[1] = 34.7;
sdr.Compute(2, classification, new int[] { 0, 6, 9, 11 });
Assert.IsTrue(sdr.weightMatrix.Matrix.Count == 7);
classification[0] = 4;
classification[1] = 34.7;
sdr.Compute(recordNumber++, classification, new int[] { 0, 6, 9, 11 });
Assert.IsTrue(sdr.weightMatrix.Matrix.Count == 7);
}
public void TestBucketEntriesSingleInput()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
List <Object> classification = new List <object>();
classification.Add(4);
classification.Add(34.7);
sdr.Compute(0, classification, new int[] { 0, 6, 9, 11 });
Assert.AreEqual(1, sdr.bucketEntries.Count);
Assert.AreEqual(34.7, sdr.bucketEntries[4][0]);
}
public void TestComputeShouldThrowObjectShouldNotBeNUllExceptionForClassification()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
try
{
sdr.Compute(0, null, null);
}
catch (ObjectShouldNotBeNUllException ex)
{
Assert.AreEqual(CLASSIFICATION_CANNOT_BE_NULL, ex.Message);
}
}
public void TestComputeSingleValueMultipleIterations()
{
double alpha = 1.0;
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>(alpha);
List <Object> classification = new List <Object>(); // contains bucket index and actual values
classification.Add(0);
classification.Add(10);
for (int i = 0; i < 10; i++)
{
sdr.Compute(i, classification, new int[] { 1, 5 });
}
Assert.AreEqual(1, sdr.Predict(new int[] { 1, 5 })[0], 0.5);
}
public void TestComputeShouldThrowObjectShouldNotBeNUllExceptionForPatternNZ()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
List <object> list = new List <object>();
try
{
sdr.Compute(0, list, null);
}
catch (ObjectShouldNotBeNUllException ex)
{
Assert.AreEqual(PATTERN_NZ_CANNOT_BE_NULL, ex.Message);
}
}
public void TestGrowingMatrixAccordingToInputs()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
List <Object> classification = new List <object>();
classification.Add(4);
classification.Add(34.7);
int recordNumber = 0;
sdr.Compute(recordNumber, classification, new int[] { 0, 6, 9, 11 });
// Doing minus one since in the code we are actually creating a column at 11th index
Assert.AreEqual(11, sdr.weightMatrix.Matrix[0].Count - 1);
sdr.Compute(recordNumber, classification, new int[] { 0, 6, 9, 5 });
// No change since the maximum input value remains 11 only
Assert.AreEqual(11, sdr.weightMatrix.Matrix[0].Count - 1);
}
public void TestPredictShouldThrowEmptyBucketException()
{
double alpha = 0.1;
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>(alpha);
List <Object> classification = new List <object>();
classification.Add(4);
classification.Add(34.7);
try
{
sdr.Predict(new int[] { 1, 5 });
}
catch (Exception ex)
{
Assert.AreEqual(ex.Message, EMPTY_BUCKET_EXCEPTION);
}
}
public void TestComputeSingleValueOneIteration()
{
double alpha = 0.1;
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>(alpha);
List <Object> classification = new List <object>();
classification.Add(4);
classification.Add(34.7);
// After this step bucket values will be updated
sdr.Compute(0, classification, new int[] { 1, 5 });
Assert.AreEqual(34.7, sdr.bucketEntries[4][0]);
double[] predictionResult = sdr.Predict(new int[] { 1, 5 });
// NOTE: if you take some of all predictions it should round upto ~ 1
Assert.AreEqual(0.19151941467016101, predictionResult[0]);
Assert.AreEqual(0.19151941467016101, predictionResult[1]);
Assert.AreEqual(0.19151941467016101, predictionResult[2]);
Assert.AreEqual(0.19151941467016101, predictionResult[3]);
// Implies bucket 4 has highest probability to come
Assert.AreEqual(0.23392234131935596, predictionResult[4]);
}
public void TestComplexLearning()
{
double alpha = 0.1;
int recordNumber = 1;
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>(alpha);
List <object> classification = new List <object>(2);
classification.Add(4);
classification.Add(34.7);
sdr.Compute(recordNumber, classification, new int[] { 1, 5, 9 });
recordNumber += 1;
classification[0] = 5;
classification[1] = 41.7;
sdr.Compute(recordNumber, classification, new int[] { 0, 6, 9, 11 });
recordNumber += 1;
classification[0] = 5;
classification[1] = 44.9;
sdr.Compute(recordNumber, classification, new int[] { 6, 9 });
recordNumber += 1;
classification[0] = 4;
classification[1] = 42.9;
sdr.Compute(recordNumber, classification, new int[] { 1, 5, 9 });
classification[0] = 4;
classification[1] = 34.7;
double[] predictedValues = sdr.Predict(new int[] { 1, 5, 9 });
Assert.AreEqual(0.11184534926411586, predictedValues[0]);
Assert.AreEqual(0.11184534926411586, predictedValues[1]);
Assert.AreEqual(0.11184534926411586, predictedValues[2]);
Assert.AreEqual(0.11184534926411586, predictedValues[3]);
Assert.AreEqual(0.30775551952585922, predictedValues[4]);
Assert.AreEqual(0.24486308341767732, predictedValues[5]);
}
public void TestDefaultConstructor()
{
SdrClassifier <int, int> sdr = new SdrClassifier <int, int>();
Assert.IsNotNull(sdr);
}
