def test_dataset(self):
dataset = TUHEEGAbnormalDataset(self.PATH)
preprocessing = Pipeline([
CommonChannelSet(),
LowestFrequency(),
BandPassFrequency(0.1, 47),
ToDataframe(),
DynamicWindow(4),
JoinedPreprocessor(inputs=[[
BinarizedSpearmanCorrelation(),
CorrelationToAdjacency()
],
Bandpower()],
output=GraphWithFeatures())
])
dataset = dataset.set_pipeline(preprocessing).load()
BinarizedSpearmanCorrelation, ToNumpy
dataset = TUHEEGAbnormalDataset('../../data/tuh_eeg_abnormal/v2.0.0/edf')
dataset.set_cache_manager(SinglePickleCache('../../export'))
preprocessing = Pipeline([
CommonChannelSet(),
LowestFrequency(),
BandPassFrequency(0.1, 47),
ToDataframe(),
DynamicWindow(8),
BinarizedSpearmanCorrelation(),
ToNumpy()
])
dataset = dataset.set_pipeline(preprocessing).load()
data, labels = dataset['data'], dataset['labels']
adjs = data[0].shape[0]
classes = len(set(labels))
input_shape = data[0].shape[1:]
inputs = [[] for _ in range(adjs)]
for d in data:
for i in range(adjs):
inputs[i].append(d[i].reshape((*input_shape, 1)))
data = [np.array(i) for i in inputs]
total_acc = 0
n_splits = 10
skf = StratifiedKFold(n_splits=n_splits, shuffle=True)