def select_active(self, domain, data, formula, active_indices):
all_violations = list(
AllViolationsStrategy.select_active(self, domain, data, formula,
active_indices))
self.last_violations = all_violations
sample_size = min(self.sample_size, len(all_violations))
import sampling
return sampling.sample_weighted(
zip(all_violations, [self.weights[i] for i in all_violations]),
sample_size)
def get_initial_indices(self):
if self.initial is None:
return list(range(len(self.data)))
elif self.initial == "random":
return random.sample(range(len(self.data)), self.initial_size)
elif self.initial == "dt_weighted":
import sampling
return sampling.sample_weighted(zip(range(len(self.data)), self.get_dt_weights()), self.initial_size)
else:
raise RuntimeError("Unknown initial type {}".format(self.initial))
def select_active(
self, domain, data, labels, formula,
active_indices) -> Tuple[np.ndarray, np.ndarray, List[int]]:
data, labels, all_violations = AllViolationsStrategy.select_active(
self, domain, data, labels, formula, active_indices)
self.last_violations = list(all_violations)
sample_size = min(self.sample_size, len(self.last_violations))
import sampling
return data, labels, sampling.sample_weighted(
zip(self.last_violations,
[self.weights[i] for i in self.last_violations]), sample_size)