Skip to content

added balancing flag to all splitting methods #328

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ymahlich merged 1 commit into from
Mar 5, 2025
Merged

added balancing flag to all splitting methods #328

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 37 additions & 7 deletions coderdata/dataset/dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -343,6 +343,7 @@ def split_train_other(
]='mixed-set',
ratio: tuple[int, int, int]=(8,2),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
) -> TwoWaySplit:
Expand All @@ -352,6 +353,7 @@ def split_train_other(
split_type=split_type,
ration=ratio,
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
**kwargs
)
Expand All @@ -366,15 +368,16 @@ def split_train_test_validate(
]='mixed-set',
ratio: tuple[int, int, int]=(8,1,1),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
) -> Split:

split = split_train_test_validate(
data=self,
split_type=split_type,
ratio=ratio,
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
**kwargs
)
Expand All @@ -389,6 +392,7 @@ def train_test_validate(
]='mixed-set',
ratio: tuple[int, int, int]=(8,1,1),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
) -> Split:
Expand All @@ -398,6 +402,7 @@ def train_test_validate(
split_type=split_type,
ratio=ratio,
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
**kwargs
)
Expand Down Expand Up @@ -725,6 +730,7 @@ def split_train_other(
]='mixed-set',
ratio: tuple[int, int, int]=(8,2),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
):
Expand All @@ -733,8 +739,9 @@ def split_train_other(
split_type,
ratio,
stratify_by,
balance,
random_state,
kwargs=kwargs
**kwargs
)
if stratify_by is not None:
train.experiments = train.experiments[train.experiments['dose_response_metric'] != 'split_class']
Expand All @@ -749,10 +756,10 @@ def split_train_test_validate(
]='mixed-set',
ratio: tuple[int, int, int]=(8,1,1),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
) -> Split:

# Type checking split_type
if split_type not in [
'mixed-set', 'drug-blind', 'cancer-blind'
Expand All @@ -766,17 +773,19 @@ def split_train_test_validate(
split_type=split_type,
ratio=[ratio[0], ratio[1] + ratio[2]],
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
kwargs=kwargs,
**kwargs,
)

test, val = _split_two_way(
data=other,
split_type=split_type,
ratio=[ratio[1], ratio[2]],
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
kwargs=kwargs,
**kwargs,
)

if stratify_by is not None:
Expand All @@ -794,6 +803,7 @@ def train_test_validate(
]='mixed-set',
ratio: tuple[int, int, int]=(8,1,1),
stratify_by: Optional[str]=None,
balance: bool=False,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
) -> Split:
Expand Down Expand Up @@ -871,8 +881,9 @@ def train_test_validate(
split_type=split_type,
ratio=ratio,
stratify_by=stratify_by,
balance=balance,
random_state=random_state,
kwargs=kwargs,
**kwargs,
)


Expand Down Expand Up @@ -976,6 +987,20 @@ def _filter(data: Dataset, split: pd.DataFrame) -> Dataset:

return data_ret

def _balance_data(
data: pd.Dataframe,
random_state: Optional[Union[int,RandomState]]=None,
# oversample: bool=False,
) -> pd.Dataframe:
tmp = deepcopy(data)
counts = tmp.value_counts('split_class')
ret_df = (
tmp
.groupby('split_class')
.sample(n=min(counts), random_state=random_state)
)
return ret_df


def _create_classes(
data: pd.DataFrame,
Expand Down Expand Up @@ -1072,6 +1097,7 @@ def _split_two_way(
'mixed-set', 'drug-blind', 'cancer-blind'
]='mixed-set',
ratio: tuple[int, int, int]=(8,2),
balance: bool=False,
stratify_by: Optional[str]=None,
random_state: Optional[Union[int,RandomState]]=None,
**kwargs: dict,
Expand Down Expand Up @@ -1150,7 +1176,6 @@ def _split_two_way(
thresh = kwargs.get('thresh', None)
num_classes = kwargs.get('num_classes', 2)
quantiles = kwargs.get('quantiles', True)

# Type checking split_type
if split_type not in [
'mixed-set', 'drug-blind', 'cancer-blind'
Expand Down Expand Up @@ -1290,6 +1315,11 @@ def _split_two_way(
thresh=thresh,
quantiles=quantiles,
)
if balance:
df_full = _balance_data(
data=df_full,
random_state=random_state
)
if split_type == 'mixed-set':
# Using StratifiedShuffleSplit to generate randomized train
# and 'other' set, since there is no need for grouping.
Expand Down