def red_wine_quality(test_size=0.2):
data = np.loadtxt("data/regression/wine_quality/winequality-red.csv",
delimiter=';',
skiprows=1)
x, y = data[:, :11], data[:, 11]
return data_util.normalize_split(x, y, test_size)
def facebook_metric(test_size=0.2):
data = np.loadtxt("data/regression/facebook_metrics/dataset_Facebook.csv",
delimiter=';',
skiprows=1,
converters={1: post_type})
x, y = data[:, :18], data[:, 18]
return data_util.normalize_split(x, y, test_size)
def breast_cancer(test_size=0.2):
data = np.loadtxt("data/classification/breast_cancer/wdbc.data",
delimiter=',',
skiprows=0,
converters={1: cancer_type_num})
x, y = data[:, 2:32], data[:, 1]
return data_util.normalize_split(x, y, test_size)
def community_crime(test_size=0.2):
used_cols = set(range(0, 128)) - set(range(0, 5)) - set(range(
101, 118)) - set(range(121, 127))
data = np.loadtxt("data/regression/communities.data",
delimiter=',',
usecols=used_cols)
x, y = data[:, :99], data[:, 99]
return data_util.normalize_split(x, y, test_size)
def gpu(test_size=0.2):
data = pd.read_csv('data/regression/gpu/sgemm_product.csv')
# from the 4 runs, use the average as target
data['average'] = data.iloc[:, 14:18].median(axis=1)
# take logarithm as suggested in data set readme
data.average = data.average.apply(np.log10)
x, y = data.iloc[:, :14], data.iloc[:, 18]
return data_util.normalize_split(x, y, test_size)
def yeast(test_size=0.2):
data = pd.read_csv('data/classification/yeast/yeast.data',
delimiter='\s+',
names=range(10))
# encode target classes as numbers
enc = preprocessing.OrdinalEncoder()
enc.fit(data.select_dtypes(include=object))
x = data.iloc[:, 1:9]
y = enc.transform(data.select_dtypes(include=object))
return data_util.normalize_split(x, y[:, 1], test_size)
def seismic_bumps(test_size=0.2):
data = np.loadtxt("data/classification/seismic-bumps.arff",
delimiter=',',
skiprows=155,
converters={
0: seismic_level,
1: seismic_level,
2: shift,
7: seismic_level
})
x, y = data[:, :18].astype(int), data[:, 18].astype(int)
return data_util.normalize_split(x, y, test_size)
def molecular(test_size=0.2):
file = 'data/regression/molecular/ACT4_competition_training.csv'
with open(file) as f:
cols = f.readline().rstrip('\n').split(',')
X = np.loadtxt(file,
delimiter=',',
usecols=range(2, len(cols)),
skiprows=1,
dtype=np.uint8)
y = np.loadtxt(file, delimiter=',', usecols=[1], skiprows=1)
np.savez('act4.npz', X, y)
file = 'data/regression/molecular/ACT2_competition_training.csv'
with open(file) as f:
cols = f.readline().rstrip('\n').split(',')
X = np.loadtxt(file,
delimiter=',',
usecols=range(2, len(cols)),
skiprows=1,
dtype=np.uint8)
y = np.loadtxt(file, delimiter=',', usecols=[1], skiprows=1)
np.savez('act2.npz', X, y)
ac4 = np.load('act4.npz')
ac2 = np.load('act2.npz')
x4, y4 = ac4['arr_0'], ac4['arr_1']
x2, y2 = ac2['arr_0'], ac2['arr_1']
x4_train, x4_test, y4_train, y4_test = data_util.normalize_split(
x4, y4, test_size)
x2_train, x2_test, y2_train, y2_test = data_util.normalize_split(
x2, y2, test_size)
return x4_train, x4_test, y4_train, y4_test, x2_train, x2_test, y2_train, y2_test
def plates(test_size=0.2):
data = np.loadtxt('data/classification/Plates/Faults.NNA', delimiter='\t')
x, y = data[:, :27], data[:, 27:]
_y, y_class = np.nonzero(y) # simple coding of target data
return data_util.normalize_split(x, y_class, test_size)
def QSAR(test_size=0.2):
data = np.loadtxt("data/regression/qsar_aquatic_toxicity.csv",
delimiter=';')
x, y = data[:, :9], data[:, 9]
return data_util.normalize_split(x, y, test_size)
def concrete(test_size=0.2):
data = pd.read_excel('data/regression/concrete/Concrete_Data.xls')
x, y = data.iloc[:, :8], data.iloc[:, 8]
return data_util.normalize_split(x, y, test_size)
def credit_card_client(test_size=0.2):
data = np.loadtxt("data/classification/default_credit_card_clients.csv",
delimiter=',',
skiprows=2)
x, y = data[:, 1:24], data[:, 24]
return data_util.normalize_split(x, y, test_size)
def statlog_australian(test_size=0.2):
data = np.loadtxt("data/classification/australian.dat", skiprows=0)
x, y = data[:, :14], data[:, 14]
return data_util.normalize_split(x, y, test_size)
def statlog_german(test_size=0.2):
data = np.loadtxt("data/classification/german.data-numeric", skiprows=0)
x, y = data[:, :24].astype(int), data[:, 24].astype(int)
return data_util.normalize_split(x, y, test_size)
def diabetic_retinopathy(test_size=0.2):
data = np.loadtxt("data/classification/messidor_features.arff",
delimiter=',',
skiprows=24)
x, y = data[:, :18], data[:, 18]
return data_util.normalize_split(x, y, test_size)