def import_dataset_arff(f, explain_indices: List[int],
random_explain_dataset: bool) -> Tuple[Dataset, Dataset, List[str]]:
dataset = load_arff(f)
dataset_len = len(dataset)
train_indices = list(range(dataset_len))
if random_explain_dataset:
random.seed(1)
# small dataset
MAX_SAMPLE_COUNT = 100
if dataset_len < (2 * MAX_SAMPLE_COUNT):
samples = int(0.2 * dataset_len)
else:
samples = MAX_SAMPLE_COUNT
# Randomly pick some instances to remove from the training dataset and use in the
# explain dataset
explain_indices = list(random.sample(train_indices, samples))
for i in explain_indices:
train_indices.remove(i)
train_dataset = Dataset.from_indices(train_indices, dataset)
explain_dataset = Dataset.from_indices(explain_indices, dataset)
return train_dataset, explain_dataset, [str(i) for i in explain_indices]
def import_datasets_arff(f, f_explain, explain_indices: List[int],
random_explain_dataset: bool) -> Tuple[Dataset, Dataset, List[str]]:
train_dataset = load_arff(f)
explain_dataset = load_arff(f_explain)
len_explain_dataset = len(explain_dataset)
if random_explain_dataset:
random.seed(7)
explain_indices = list(random.sample(range(len_explain_dataset), 300))
explain_dataset = Dataset.from_indices(explain_indices, explain_dataset)
return train_dataset, explain_dataset, [str(i) for i in explain_indices]
