def add_sample_to_set(self, sample_id: Sample, sample_set: SplitSampleSet):
"""
Add new sample to sample set.
"""
split = sample_set.split_type
self.sample_splits[sample_id] = split
self.unused_sample_ids.remove(sample_id)
sample_set.update(sample_id, self.full_sample_set, inplace=True)
def peek_sample(self, sample_id: str, sample_set_to_update: SplitSampleSet,
other_sample_set: SplitSampleSet) -> float:
"""
Check score for adding sample `sample_id` to `sample_set_to_update` without actually
making the update (not in-place).
Args:
sample_id (str): id of sample to check update for.
sample_set_to_update (SplitSampleSet): Sample set to be updated with chosen sample.
other_sample_set (SplitSampleSet): The other sample set (not updated).
Returns:
float: Split score if we had added `sample_id` to `sample_set_to_update`
"""
a_dist, c_dist = sample_set_to_update.update(sample_id, self.full_sample_set,
inplace=False)
if sample_set_to_update.is_train:
train_a_dist = a_dist
train_c_dist = c_dist
test_a_dist = other_sample_set.atom_distribution
test_c_dist = other_sample_set.compound_distribution
else:
test_a_dist = a_dist
test_c_dist = c_dist
train_a_dist = other_sample_set.atom_distribution
train_c_dist = other_sample_set.compound_distribution
return self.score(train_a_dist, test_a_dist, train_c_dist, test_c_dist)
def peek_ray(sample_id: str, sample_set_to_update: SplitSampleSet,
other_sample_set: SplitSampleSet,
dbca_config: DBCASplitterConfig, full_sample_set: FullSampleSet) -> float:
"""
"""
a_dist, c_dist = sample_set_to_update.update(sample_id,
full_sample_set,
inplace=False)
if sample_set_to_update.is_train:
train_a_dist = a_dist
train_c_dist = c_dist
test_a_dist = other_sample_set.atom_distribution
test_c_dist = other_sample_set.compound_distribution
else:
test_a_dist = a_dist
test_c_dist = c_dist
train_a_dist = other_sample_set.atom_distribution
train_c_dist = other_sample_set.compound_distribution
return score(train_a_dist, test_a_dist, train_c_dist, test_c_dist,
dbca_config)
def peek_ray(sample_id: str, sample_set_to_update: SplitSampleSet,
other_sample_set: SplitSampleSet, dbca_config: DBCASplitterConfig,
full_sample_set: FullSampleSet) -> float:
"""
"""
# print(f"[_outer_peek_mp]: Starting work on {sample_id}... ")
a_dist, c_dist = sample_set_to_update.update(sample_id,
full_sample_set,
inplace=False)
if sample_set_to_update.is_train:
train_a_dist = a_dist
train_c_dist = c_dist
test_a_dist = other_sample_set.atom_distribution
test_c_dist = other_sample_set.compound_distribution
else:
test_a_dist = a_dist
test_c_dist = c_dist
train_a_dist = other_sample_set.atom_distribution
train_c_dist = other_sample_set.compound_distribution
# print(f"[_outer_peek_mp]: Done work on {sample_id}...! ")
return score(train_a_dist, test_a_dist, train_c_dist, test_c_dist,
dbca_config)