def test_sanity_checks():
"""Ensure that sanity checks are performed, and as expected."""
### -------------- as you add them to dataset --------------
with raises(EmptyFeatureSetException):
ds.add_samplet('empty_features', [], 'target')
### -------------- as you save them to disk --------------
ds.add_samplet('all_zeros', np.zeros((ds.num_features, 1)), 'target')
with raises(ConstantValuesException):
ds.save(out_file)
ds.del_samplet('all_zeros')
# checking for random constant value!
const_value = np.random.randint(10, 100)
const_feat_set = np.full((ds.num_features, 1), const_value)
ds.add_samplet('all_constant', const_feat_set, 'target')
with raises(ConstantValuesException):
ds.save(out_file)
# now checking for constants across samplets
# this is easily achieved by adding different samplets with same features
# such a bug is possible, when user made a mistake querying
# the right files for the right samplet ID
const_ds = ClfDataset()
rand_feat_same_across_samplets = np.random.randn(10)
for index in range(np.random.randint(10, 100)):
const_ds.add_samplet(str(index), rand_feat_same_across_samplets, index)
with raises(ConstantValuesException):
const_ds.save(out_file)
def make_random_Dataset(max_num_classes=20,
max_class_size=50,
max_dim=100,
stratified=True):
"Generates a random Dataset for use in testing."
smallest = 10
max_class_size = max(smallest, max_class_size)
largest = max(50, max_class_size)
largest = max(smallest + 3, largest)
if max_num_classes != 2:
num_classes = np.random.randint(2, max_num_classes, 1)
else:
num_classes = 2
if type(num_classes) == np.ndarray:
num_classes = num_classes[0]
if not stratified:
class_sizes = np.random.random_integers(smallest,
largest,
size=[num_classes, 1])
else:
class_sizes = np.repeat(np.random.randint(smallest, largest),
num_classes)
num_features = np.random.randint(min(3, max_dim), max(3, max_dim), 1)[0]
feat_names = [str(x) for x in range(num_features)]
class_ids = list()
labels = list()
for cl in range(num_classes):
class_ids.append('class-{}'.format(cl))
labels.append(int(cl))
ds = ClfDataset()
for cc, class_ in enumerate(class_ids):
subids = [
'sub{:03}-class{:03}'.format(ix, cc)
for ix in range(class_sizes[cc])
]
for sid in subids:
ds.add_samplet(samplet_id=sid,
features=feat_generator(num_features),
target=class_,
feature_names=feat_names)
return ds
def make_fully_separable_classes(max_class_size=50, max_dim=100):
from sklearn.datasets import make_blobs
random_center = np.random.rand(max_dim)
cluster_std = 1.5
centers = [random_center, random_center + cluster_std * 6]
blobs_X, blobs_y = make_blobs(n_samples=max_class_size, n_features=max_dim,
centers=centers, cluster_std=cluster_std)
unique_labels = np.unique(blobs_y)
class_ids = {lbl: str(lbl) for lbl in unique_labels}
new_ds = ClfDataset()
for index, row in enumerate(blobs_X):
new_ds.add_samplet(samplet_id='sub{}'.format(index),
features=row, # label=blobs_y[index],
target=class_ids[blobs_y[index]])
return new_ds
def load_arff_dataset(ds_path):
"""Convenience utility to quickly load ARFF files into pyradigm format"""
try:
ds = ClassificationDataset.from_arff(ds_path)
except:
try:
ds = RegressionDataset.from_arff(ds_path)
except:
try:
ds = MLDataset(arff_path=ds_path)
except:
raise TypeError(
'Error in loading the ARFF dataset @ path below!'
' Ignoring {}'.format(ds_path))
return ds
def load_dataset(ds_path):
"""Convenience utility to quickly load any type of pyradigm dataset"""
try:
ds = ClassificationDataset(dataset_path=ds_path)
except:
try:
ds = RegressionDataset(dataset_path=ds_path)
except:
try:
warn(
'MLDtaset is deprecated. Switch to the latest pyradigm data '
'structures such as ClassificationDataset or '
'RegressionDataset as soon as possible.')
ds = MLDataset(filepath=ds_path)
except:
raise TypeError('Dataset class @ path below not recognized!'
' Must be a valid instance of one of '
'ClassificationDataset or '
'RegressionDataset or MLDataset.\n'
' Ignoring {}'.format(ds_path))
return ds
def get_features(samplet_id_list,
classes,
featdir,
outdir,
outname,
get_method=None,
feature_type='dir_of_dris'):
"""
Populates the pyradigm data structure with features from a given method.
Parameters
----------
samplet_id_list : list or ndarray
List of subject IDs
classes : dict
dict of class labels keyed in by subject id
featdir : str
Path to input directory to read the features from
outdir : str
Path to output directory to save the gathered features to.
outname : str
Name of the feature set
get_method : callable
Callable that takes in a path and returns a vectorized feature set
e.g. set of subcortical volumes, with an optional array of names for each
feature.
feature_type : str
Identifier of data organization for features.
Returns
-------
saved_path : str
Path where the features have been saved to as a pyradigm dataset
"""
if not callable(get_method):
raise ValueError("Supplied get_method is not callable! "
" It must take in a path and "
"return a vectorized feature set and labels.")
# generating an unique numeric label for each class
# (sorted in order of their appearance in metadata file)
class_set = set(classes.values())
class_labels = dict()
for idx, cls in enumerate(class_set):
class_labels[cls] = idx
ids_excluded = list()
if feature_type == 'data_matrix':
data_matrix = get_data_matrix(featdir)
ds = ClassificationDataset()
for samplet_id in samplet_id_list:
try:
if feature_type == 'data_matrix':
data = data_matrix[samplet_id_list.index(samplet_id), :]
feat_names = None
else:
data, feat_names = get_method(featdir, samplet_id)
ds.add_samplet(samplet_id=samplet_id,
features=data,
target=classes[samplet_id],
feature_names=feat_names)
except:
ids_excluded.append(samplet_id)
traceback.print_exc()
warnings.warn(
"Features for {} via {} method could not be read or added."
" Excluding it.".format(samplet_id, get_method.__name__))
# warning for if failed to extract features even for one subject
alert_failed_feature_extraction(len(ids_excluded), ds.num_samplets,
len(samplet_id_list))
# save the dataset to disk to enable passing on multiple dataset(s)
saved_path = realpath(pjoin(outdir, outname))
try:
ds.save(saved_path)
except IOError as ioe:
print('Unable to save {} features to disk in folder:\n{}'
''.format(outname, outdir))
raise ioe
return saved_path
estimator = 'randomforestclassifier' # 'svm' #
dr_method = 'isomap' # 'selectkbest_f_classif' # 'variancethreshold' #
dr_size = 'tenth'
gs_level = 'none' # 'light'
random.seed(42) # to save time for local tests
covar_list = ('age', 'gender', 'dummy')
covar_types = ('age', 'gender', 'float')
covar_arg = ' '.join(['age', 'gender'])
deconf_method = 'residualize'
out_path1 = os.path.join(out_dir, 'random_clf_ds1.pkl')
out_path2 = os.path.join(out_dir, 'random_clf_ds2.pkl')
if pexists(out_path1) and pexists(out_path2):
ds_one = ClassificationDataset(dataset_path=out_path1)
ds_two = ClassificationDataset(dataset_path=out_path2)
else:
ds_one = make_random_ClfDataset(max_num_classes=max_num_classes,
stratified=True,
max_class_size=max_class_size,
max_dim=max_dim,
min_num_classes=min_num_classes,
attr_names=covar_list,
attr_types=covar_types)
ds_one.save(out_path1)
ds_two = dataset_with_new_features_same_everything_else(ds_one, max_dim)
ds_two.save(out_path2)
A = 0