def stopRecording(self):
self.recording = False
print(self.eeg_raw)
eeg_epochs = BCI.epoch_array(self.eeg_raw,
self.EPOCH_LENGTH,
self.OVERLAP_LENGTH * self.freq,
self.freq)
print(eeg_epochs.shape)
feat_matrix = BCI.compute_feature_matrix(eeg_epochs,
self.freq)
percent_change = BCI.calc_ratio(feat_matrix, self.baseline)
print(percent_change)
q75, q25 = np.percentile(percent_change, [75, 25])
iqr = q75 - q25
lower_bound = q25 - (1.5 * iqr)
upper_bound = q75 + (1.5 * iqr)
for x in range(0,len(percent_change)):
if (percent_change[x] > upper_bound):
percent_change[x] = np.median(percent_change)
if (percent_change[x] < lower_bound):
percent_change[x] = np.median(percent_change)
# return percent_change
# data = self.processEEG(self.eeg_raw)
self.eeg_raw = np.ndarray(shape = (8,4))
return jsonify(percent_change)
def __init__(self):
print('Connecting...')
streams = resolve_byprop('type', 'EEG', timeout=2)
if len(streams) == 0:
raise RuntimeError('Can\'t find EEG stream.')
# set up Inlet
inlet = StreamInlet(streams[0], max_chunklen=12)
eeg_time_correction = inlet.time_correction()
# Pull relevant information
info = inlet.info()
self.desc = info.desc()
self.freq = int(info.nominal_srate())
## TRAIN DATASET
print('Recording Baseline')
eeg_data_baseline = BCI.record_eeg_filtered(
self.TRAINING_LENGTH,
self.freq,
self.INDEX_CHANNEL,
True, )
eeg_epochs_baseline = BCI.epoch_array(
eeg_data_baseline,
self.EPOCH_LENGTH,
self.OVERLAP_LENGTH * self.freq,
self.freq)
feat_matrix_baseline = BCI.compute_feature_matrix(
eeg_epochs_baseline,
self.freq)
self.baseline = BCI.calc_baseline(feat_matrix_baseline)
def processEEG(eeg_raw):
eeg_epochs = BCI.epoch_array(eeg_raw, self.EPOCH_LENGTH,
self.OVERLAP_LENGTH * self.freq,
self.freq)
feat_matrix = BCI.compute_feature_matrix(eeg_epochs, self.freq)
percent_change = BCI.calc_ratio(feat_matrix, self.baseline)
q75, q25 = np.percentile(percent_change, [75, 25])
iqr = q75 - q25
lower_bound = q25 - (1.5 * iqr)
upper_bound = q75 + (1.5 * iqr)
for x in range(0, len(percent_change)):
if (percent_change[x] > upper_bound):
percent_change[x] = np.median(percent_change)
if (percent_change[x] < lower_bound):
percent_change[x] = np.median(percent_change)
return percent_change
print("Move Eyes Up and Down")
eeg_data2 = BCI.record_eeg(TRAINING_LENGTH, freq, INDEX_CHANNEL)
print("Blink Rapidly ")
eeg_data3 = BCI.record_eeg(TRAINING_LENGTH, freq, INDEX_CHANNEL)
# Divide data into epochs
eeg_epochs0 = BCI.epoch_array(eeg_data0, EPOCH_LENGTH, OVERLAP_LENGTH * freq, freq)
eeg_epochs1 = BCI.epoch_array(eeg_data0, EPOCH_LENGTH, OVERLAP_LENGTH * freq, freq)
eeg_epochs2 = BCI.epoch_array(eeg_data0, EPOCH_LENGTH, OVERLAP_LENGTH * freq, freq)
eeg_epochs3 = BCI.epoch_array(eeg_data0, EPOCH_LENGTH, OVERLAP_LENGTH * freq, freq)
# Computer corresponding features
feat_matrix0 = BCI.compute_feature_matrix(eeg_epochs0, freq)
feat_matrix1 = BCI.compute_feature_matrix(eeg_epochs1, freq)
feat_matrix2 = BCI.compute_feature_matrix(eeg_epochs2, freq)
feat_matrix3 = BCI.compute_feature_matrix(eeg_epochs3, freq)
# Train Classifier
[classifier, mu_ft, std_ft, score] = BCI.train_classifier(
feat_matrix0,
feat_matrix1,
feat_matrix2,
feat_matrix3,
'RandomForestClassifier')
print(str(score * 100) + '% correctly predicted')
freq,
INDEX_CHANNEL,
True,
)
print('Recording Main')
eeg_data_main = BCI.record_eeg_filtered(TRAINING_LENGTH, freq,
INDEX_CHANNEL)
# Divide data into epochs
eeg_epochs_baseline = BCI.epoch_array(eeg_data_baseline, EPOCH_LENGTH,
OVERLAP_LENGTH * freq, freq)
eeg_epochs_main = BCI.epoch_array(eeg_data_main, EPOCH_LENGTH,
OVERLAP_LENGTH * freq, freq)
# Computer corresponding features
feat_matrix_baseline = BCI.compute_feature_matrix(eeg_epochs_baseline,
freq)
feat_matrix_main = BCI.compute_feature_matrix(eeg_epochs_main, freq)
baseline = BCI.calc_baseline(feat_matrix_baseline)
percent_change = BCI.calc_ratio(feat_matrix_main, baseline)
q75, q25 = np.percentile(percent_change, [75, 25])
iqr = q75 - q25
lower_bound = q25 - (1.5 * iqr)
upper_bound = q75 + (1.5 * iqr)
print(iqr)
print(percent_change)
print(baseline)
for x in range(0, len(percent_change)):
if (percent_change[x] > upper_bound):
percent_change[x] = np.median(percent_change)