def dataSetTestATest(cov, means):
svg = ClassificationValuesGenerator(0, 30)
return svg.getSyntheticValuesForClassificationWithMeans(
[50] * results.numberOfClasses, cov, means)
parser.add_argument(
"-e",
action="store",
dest="testUsingTrainingData",
type=int,
default=1,
help=
"1: Test de classifier using a different data set. 0: Test using the training data set."
)
if __name__ == "__main__":
results = parser.parse_args()
if results.numberOfClasses > 1:
numberOfDataPerClass = np.random.uniform(80, 100,
results.numberOfClasses)
svg = ClassificationValuesGenerator(0, 30)
values = svg.getSyntheticValuesForClassification(numberOfDataPerClass)
trainingData = values[0]
cov = values[1]
means = values[2]
classificator = LinearClassificator()
classificator.findW(trainingData)
classificable = []
classificated = [[] for i in range(0, results.numberOfClasses)]
if results.testUsingTrainingData == 0:
testData = trainingData
else:
def dataSetTestATraining():
numberOfDataPerClass = np.random.uniform(80, 100, results.numberOfClasses)
svg = ClassificationValuesGenerator(0, 30)
return svg.getSyntheticValuesForClassification(numberOfDataPerClass)
def dataSetTestATrainingWithFixedDistribution():
numberOfDataPerClass = np.random.uniform(80, 100, 2)
svg = ClassificationValuesGenerator()
cov = np.array([[1, 0], [0, 1]])
means = [[0, 0], [0, 4]]
return [svg.getSyntheticValuesForClassificationWithMeans(numberOfDataPerClass, cov, means), cov, means]
def dataSetTestATest(cov, means):
svg = ClassificationValuesGenerator(0, 1)
return svg.getSyntheticValuesForClassificationWithMeans([50] * 2, cov, means)
def dataSetTestBTraining(dim):
numberOfDataPerClass = np.random.uniform(80, 100, 3)
svg = ClassificationValuesGenerator(0, 10)
return svg.getSyntheticValuesForClassification(numberOfDataPerClass, dim)
parser = argparse.ArgumentParser(description="Mixture of Gaussians of K classes with D = 2.")
parser.add_argument("-k", action="store", dest="numberOfClasses", type=int, default=2,
help="Number of classses.")
parser.add_argument("-t", action="store", dest="testsToRun", type=int, default=0,
help="0: run EM and K-Means. 1: Will run only EM. 2: Will run only K-Means.")
parser.add_argument("-em", action="store", dest="numberOfIterationsOfEM", type=int, default=2000,
help="Number of iterations of EM.")
parser.add_argument("-km", action="store", dest="numberOfIterationsOfKMeans", type=int, default=200,
help="Number of iterations of K-Means.")
if __name__ == "__main__":
results = parser.parse_args()
if results.numberOfClasses > 1:
numberOfDataPerClass = np.random.uniform(80, 100, results.numberOfClasses)
svg = ClassificationValuesGenerator(0, 10)
values = svg.getSyntheticValuesForClassification(numberOfDataPerClass)
trainingData = values[0]
cov = values[1]
means = values[2]
classificable = []
for cl in trainingData:
classificable.extend(cl)
shuffle(classificable)
if results.testsToRun == 0 or results.testsToRun == 1:
classificator1 = EM()
classificator1.expectationMaximization(classificable, results.numberOfClasses, results.numberOfIterationsOfEM)