import pyarrow as pa
from sklearn.preprocessing import Binarizer as BinarizerOperation
from DashAI.back.converters.category.encoding import EncodingConverter
from DashAI.back.converters.sklearn_wrapper import SklearnWrapper
from DashAI.back.core.schema_fields import bool_field, float_field, schema_field
from DashAI.back.core.schema_fields.base_schema import BaseSchema
from DashAI.back.core.utils import MultilingualString
from DashAI.back.types.dashai_data_type import DashAIDataType
from DashAI.back.types.value_types import Integer
class BinarizerSchema(BaseSchema):
threshold: schema_field(
float_field(),
0.0,
description=MultilingualString(
en=(
"Feature values below or equal to this are replaced by 0, "
"above it by 1."
),
es=(
"Los valores por debajo o igual al umbral se reemplazan por 0; "
"los superiores por 1."
),
),
) # type: ignore
use_copy: schema_field(
bool_field(),
True,
description=MultilingualString(
en="Set to False to perform inplace binarization.",
es="Ponlo en False para binarizar in situ.",
),
alias=MultilingualString(en="copy", es="copiar"),
) # type: ignore
[docs]
class Binarizer(EncodingConverter, SklearnWrapper, BinarizerOperation):
"""Scikit-learn's Binarizer wrapper for DashAI."""
SCHEMA = BinarizerSchema
DESCRIPTION = MultilingualString(
en=("Binarize data (set feature values to 0 or 1) according to a threshold."),
es=(
"Binariza datos (pone valores de características en 0 o 1) según un umbral."
),
)
DISPLAY_NAME = MultilingualString(en="Binarizer", es="Binarizador")
IMAGE_PREVIEW = "binarizer.png"
def get_output_type(self, column_name: str = None) -> DashAIDataType:
"""Returns Integer64 as the output type for binarized data."""
return Integer(arrow_type=pa.int64())