def convert(raw_dir, max_features):
dataset_dict = util.convert_uci_classif( info, raw_dir, file_name, y_first=True )
# Feature 15 is constant. Thus useless.
# Feature 4 is the most important feature (according to random forest).
# But with it, the problem is too easy i.e., most decent learning algorithms classify the test perfectly.
return util.remove_features( dataset_dict, [4,15] )
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
return util.convert_uci_classif(info,
raw_dir,
file_name_list,
delimiter=' ')
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
return util.convert_uci_classif(info,
raw_dir,
file_name,
converters={0: util.convert_date})
def convert(raw_dir, max_features):
util.untar(raw_dir, file_name)
file_name_list = ["census-income.data", "census-income.data"]
# We remove feature 24 (instance weight).
columns = range(24) + range(25, 42)
return util.convert_uci_classif(info, raw_dir, file_name_list, delimiter=", ", usecols=columns)
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
info['x'], info['y'] = util.convert_uci_classif(
info['x_type'], info['y_type'], raw_dir,'anneal.data')
return info
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
data_dir = os.path.join( raw_dir, 'ml-prove' )
file_name_list = ['test.csv', 'validation.csv', 'train.csv' ]
dataset_dict = util.convert_uci_classif( info, data_dir, file_name_list )
return util.remove_features(dataset_dict, [51,53])
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
data_dir = os.path.join(raw_dir, 'ml-prove')
file_name_list = ['test.csv', 'validation.csv', 'train.csv']
dataset_dict = util.convert_uci_classif(info, data_dir, file_name_list)
return util.remove_features(dataset_dict, [51, 53])
def convert(raw_dir, max_features):
util.unzip(raw_dir, file_name)
data_dir = os.path.join(os.path.join(raw_dir, "PAMAP2_Dataset"), "Protocol")
file_name_list = ["subject10%d.dat" % x for x in range(1, 10)]
# We remove features 0 (timestamp), 16-19, 33-36 and 50-53 (invalid).
columns = range(1, 16) + range(20, 33) + range(37, 50)
return util.convert_uci_classif(info, data_dir, file_name_list, stride=4, delimiter=" ", y_first=True, usecols=columns)
def convert(raw_dir, max_features):
dataset_dict = util.convert_uci_classif(info,
raw_dir,
file_name,
y_first=True)
# Feature 15 is constant. Thus useless.
# Feature 4 is the most important feature (according to random forest).
# But with it, the problem is too easy i.e., most decent learning algorithms classify the test perfectly.
return util.remove_features(dataset_dict, [4, 15])
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
columns = range(1, 32)
return util.convert_uci_classif(info,
raw_dir,
file_name,
y_first=True,
usecols=columns)
def convert(raw_dir, max_features):
# extract only one line out of 10 since the dataset fails to load in memory (using numpy.loadtxt with str data type)
file_name_ = "kddcup_sub.data"
with gzip.open( path.join(raw_dir, file_name ), 'r') as fd_read:
with open( path.join(raw_dir, file_name_), 'w') as fd_write:
for i,line in enumerate(fd_read):
if i%10 == 0:
fd_write.write( line )
return util.convert_uci_classif(info, raw_dir, file_name_)
def convert(raw_dir, max_features):
util.untar(raw_dir, file_name)
file_name_list = ["census-income.data", "census-income.data"]
# We remove feature 24 (instance weight).
columns = range(24) + range(25, 42)
return util.convert_uci_classif(info,
raw_dir,
file_name_list,
delimiter=", ",
usecols=columns)
def convert(raw_dir, max_features):
util.unzip(raw_dir, file_name)
data_dir = os.path.join(os.path.join(raw_dir, "PAMAP2_Dataset"),
"Protocol")
file_name_list = ["subject10%d.dat" % x for x in range(1, 10)]
# We remove features 0 (timestamp), 16-19, 33-36 and 50-53 (invalid).
columns = range(1, 16) + range(20, 33) + range(37, 50)
return util.convert_uci_classif(info,
data_dir,
file_name_list,
stride=4,
delimiter=" ",
y_first=True,
usecols=columns)
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
# In the data file, one instance is splitted over two lines.
# The following code write a fixed file.
fixed_file_name = file_name + '2'
with open(path.join(raw_dir,file_name)) as f_in:
with open(path.join(raw_dir,fixed_file_name),'w') as f_out:
for line in f_in.readlines():
line = line.strip()
f_out.write(line)
if len(line) > 0 and line[-1] != ',':
f_out.write('\n')
return util.convert_uci_classif(info, raw_dir, fixed_file_name)
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
# In the data file, one instance is splitted over two lines.
# The following code write a fixed file.
fixed_file_name = file_name + '2'
with open(path.join(raw_dir, file_name)) as f_in:
with open(path.join(raw_dir, fixed_file_name), 'w') as f_out:
for line in f_in.readlines():
line = line.strip()
f_out.write(line)
if len(line) > 0 and line[-1] != ',':
f_out.write('\n')
return util.convert_uci_classif(info, raw_dir, fixed_file_name)
def convert(raw_dir, max_features):
return util.convert_uci_classif(info, raw_dir, file_name)
def convert(raw_dir, max_features):
return util.convert_uci_classif(info, raw_dir, file_name)
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
return util.convert_uci_classif( info, raw_dir, file_name )
def convert(raw_dir, max_features):
"""
returns a dictionary containing the required fields for the dataset.
"""
columns = range(1,32)
return util.convert_uci_classif( info, raw_dir, file_name, y_first=True, usecols=columns )
