def main(
dataset_name="anguita2013",
fs_new=33,
win_len=2.56,
win_inc=1.0,
task="har",
split_type="predefined",
feat_name="all",
):
dataset = dataset_importer(dataset_name)
resampled = transformers.resample(parent=dataset, fs_new=fs_new)
filtered = transformers.body_grav_filter(parent=resampled)
windowed = transformers.window(parent=filtered,
win_len=win_len,
win_inc=win_inc)
accel_feats = features.ecdf(parent=windowed,
source_filter=take_all(),
source_name=feat_name,
n_components=21)
task = selectors.select_task(parent=accel_feats, task_name=task)
split = selectors.select_split(parent=task, split_type=split_type)
clf = models.sgd_classifier(parent=split,
split=split,
task=task,
data=accel_feats)
clf.evaluate_outputs()
return clf
def main(
dataset_name,
source_filter,
feat_name,
clf_name,
fs_new,
win_len,
win_inc,
split_type,
):
dataset = dataset_importer(dataset_name)
resampled = transformers.resample(
parent=dataset,
fs_new=fs_new
)
filtered = transformers.body_grav_filter(
parent=resampled
)
windowed = transformers.window(
parent=filtered,
win_len=win_len,
win_inc=win_inc
)
if feat_name == 'ecdf':
feats = features.ecdf(
parent=windowed,
source_filter=source_filter,
n_components=21
)
elif feat_name == 'statistical':
feats = features.statistical_features(
parent=windowed,
source_filter=source_filter,
)
else:
raise ValueError
task = selectors.select_task(
parent=feats,
task_name='har'
)
split = selectors.select_split(
parent=task,
split_type=split_type
)
if clf_name == 'sgd_classifier':
clf = models.sgd_classifier(
parent=split,
data=feats
)
elif clf_name == 'logistic_regression_cv':
clf = models.logistic_regression_cv(
parent=split,
data=feats
)
else:
raise ValueError
return clf.get_results()
def ecdf(name, n_components):
windowed = dataset_windowed(name)
feats = features.ecdf(parent=windowed, n_components=n_components)
return feats