def __call__(
self,
data: DataTuple,
split_id: int = 0
) -> Tuple[DataTuple, DataTuple, Dict[str, float]]:
del split_id
train_len = round(self.train_percentage * len(data))
train = data.apply_to_joined_df(
lambda df: df.iloc[:train_len].reset_index(drop=True))
train = train.replace(name=f"{data.name} - Train")
test = data.apply_to_joined_df(
lambda df: df.iloc[train_len:].reset_index(drop=True))
test = test.replace(name=f"{data.name} - Test")
assert len(train) + len(test) == len(data)
return train, test, {}
def query_dt(datatup: DataTuple, query_str: str) -> DataTuple:
"""Query a datatuple."""
assert isinstance(query_str, str)
assert isinstance(datatup, DataTuple)
def _query_func(joined_data_frame: pd.DataFrame) -> pd.DataFrame:
return dataset_from_cond(joined_data_frame, cond=query_str)
return datatup.apply_to_joined_df(_query_func)
