def predict(self, trainingResult, testData):
if self.method == 'boser':
from daal.algorithms.svm.training import boser
algorithm = prediction.Batch(method=boser, fptype=self.dtype)
if self.kernel == 'linear':
algorithm.parameter.kernel = kernel_function.linear.Batch()
elif self.kernel == 'rbf':
algorithm.parameter.kernel = kernel_function.rbf.Batch()
algorithm.input.setTable(classifier.prediction.data, testData)
algorithm.input.setModel(classifier.prediction.model,
trainingResult.get(classifier.training.model))
algorithm.compute()
predictionResult = algorithm.getResult()
predictedResponses = predictionResult.get(
classifier.prediction.prediction)
'''
block = BlockDescriptor ()
predictedResponses.getBlockOfRows (0, predictedResponses.getNumberOfRows (), readWrite, block)
predictArray = block.getArray ()
predictArray[predictArray < 0] = -1
predictArray[predictArray >= 0] = 1
predictedResponses.releaseBlockOfRows (block)
'''
return predictedResponses
def training(self, trainData, trainDependentVariables):
#Set algorithms parameters
if self.method == 'boser':
from daal.algorithms.svm.training import boser
trainingBatch = training.Batch(method=boser, fptype=self.dtype)
predictionBatch = prediction.Batch()
if self.kernel == 'linear':
trainingBatch.parameter.kernel = kernel_function.linear.Batch(
method=boser, fptype=self.dtype)
trainingBatch.parameter.k = self.k
trainingBatch.parameter.b = self.b
elif self.kernel == 'rbf':
trainingBatch.parameter.kernel = kernel_function.rbf.Batch(
method=boser, fptype=self.dtype)
trainingBatch.parameter.sigma = self.sigma
trainingBatch.parameter.cacheSize = self.cachesize
trainingBatch.parameter.C = self.C
trainingBatch.parameter.accuracyThreshold = self.tolerence
trainingBatch.parameter.tau = self.tau
trainingBatch.parameter.maxIterations = self.maxIterations
trainingBatch.parameter.doShrinking = self.doShrinking
algorithm = multi_class_classifier.training.Batch(self.classes)
algorithm.parameter.training = trainingBatch
algorithm.parameter.prediction = predictionBatch
algorithm.input.set(classifier.training.data, trainData)
algorithm.input.set(classifier.training.labels,
trainDependentVariables)
trainingResult = algorithm.compute()
return trainingResult
def predict(self, trainingResult, testData): #give other parameters
if self.method == 'boser':
from daal.algorithms.svm.training import boser
predictionBatch = prediction.Batch (method=boser, fptype=self.dtype)
trainingBatch = training.Batch (method=boser, fptype=self.dtype)
if self.kernel == 'linear':
predictionBatch.parameter.kernel = kernel_function.linear.Batch ()
elif self.kernel == 'rbf':
predictionBatch.parameter.kernel = kernel_function.rbf.Batch ()
algorithm = multi_class_classifier.prediction.Batch (self.classes)
algorithm.parameter.training = trainingBatch
algorithm.parameter.prediction = predictionBatch
algorithm.input.setTable (classifier.prediction.data, testData)
algorithm.input.setModel (classifier.prediction.model, trainingResult.get (classifier.training.model))
algorithm.compute ()
predictionResult = algorithm.getResult()
predictedResponses = predictionResult.get (classifier.prediction.prediction)
# Change the predicted values to 1 and -1
return predictedResponses
