def create_sea_drift_dataset(n_samples_per_concept=200, concepts=[0, 1, 2, 3]):
X_stream = []
Y_stream = []
concept_drifts = []
t = 0
gen = SEAGenerator()
gen.prepare_for_use()
for _ in concepts:
if t != 0:
concept_drifts.append(t)
X, y = gen.next_sample(batch_size=n_samples_per_concept)
X_stream.append(X)
Y_stream.append(y)
gen.generate_drift()
t += n_samples_per_concept
return {
"data": (np.concatenate(X_stream, axis=0),
np.concatenate(Y_stream, axis=0).reshape(-1, 1)),
"drifts":
np.array(concept_drifts)
}
accuracy_3_mv = 0.0
accuracy_3_av = 0.0
accuracy_3_goowe = 0.0
total_3 = 0.0
true_predictions_3_mv = 0.0
true_predictions_3_av = 0.0
true_predictions_3_goowe = 0.0
total = 0.
### Generating drifts by generating random values for each Stream
p1 = random.random()
p2 = random.random()
p3 = random.random()
if p1 > p1_threshold:
stream_1.generate_drift()
logging.info('\n\tDrift generatoed for STREAM 1')
if p2 > p2_threshold:
stream_2.generate_drift()
logging.info('\n\tDrift generatoed for STREAM 2')
if p3 > p3_threshold:
stream_3.generate_drift()
logging.info('\n\tDrift generatoed for STREAM 3')
total += 1
cur_1 = stream_1.next_sample()
cur_2 = stream_2.next_sample()
cur_3 = stream_3.next_sample()
X_1, y_1 = cur_1[0], cur_1[1]
X_2, y_2 = cur_2[0], cur_2[1]
X_3, y_3 = cur_3[0], cur_3[1]
preds_1 = goowe_1.predict(X_1)
