def load_sinus():
size = 1000
location = 0.0
scale = 0.5
amplitude = 0.3
freq = 3.5
x, y = DataDistributions.generateSinusCluster(size, location, scale, amplitude, freq)
x_test, y_test = DataDistributions.generateSinusCluster(size * 3, location, scale, amplitude, freq)
return x, x_test, y, y_test
def load_data(dataType):
'''
Method for loading data. Returns training and test data.
:param dataType: String for the name of the data. The data will be automatically searched in the data directory.
:return: Returns 4 arrays: x, x_test, y, y_test
'''
if dataType == "sinus":
x, x_test, y, y_test = load_sinus()
elif dataType == "cod-rna":
x, x_test, y, y_test = load_codrna()
elif dataType == "covtype":
x, x_test, y, y_test = load_covtype()
elif dataType == "banana":
x, x_test, y, y_test = load_banana()
# elif dataType == "skin":
# x, x_test, y, y_test = load_skin()
elif dataType == "cluster":
x, x_test, y, y_test = DataDistributions.generateNonLinearClusters(10000)
elif dataType == "clusterx":
x, x_test, y, y_test = DataDistributions.generateNonLinearClusters(10000, x_noise=True)
elif dataType == "clustery":
x, x_test, y, y_test = DataDistributions.generateNonLinearClusters(10000, y_noise=True)
elif dataType == "mnist":
mnist = fetch_mldata('MNIST original')
data = mnist.data
target = mnist.target
target = np.where(target > 0, 1, -1)
x, x_test, y, y_test = cv.train_test_split(data, target, train_size=0.7)
elif dataType == "shuttle":
x, x_test, y, y_test = load_libsvm_file(dataType)
y = np.where(y != 1, -1, 1)
y_test = np.where(y_test != 1, -1, 1)
else:
x, x_test, y, y_test = load_libsvm_file(dataType)
return x, x_test, y, y_test
