Skip to content

LabelEncoder

Bases: InvertibleTableTransformer

The LabelEncoder encodes one or more given columns into labels.

Parameters:

Name Type Description Default
column_names str | list[str] | None

The list of columns used to fit the transformer. If None, all non-numeric columns are used.

 None
partial_order list[Any] | None

The partial order of the labels. The labels are encoded in the order of the given list. Additional values are assigned labels in the order they are encountered during fitting.

 None
Source code in src/safeds/data/tabular/transformation/_label_encoder.py 
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
class LabelEncoder(InvertibleTableTransformer):
    """
    The LabelEncoder encodes one or more given columns into labels.

    Parameters
    ----------
    column_names:
        The list of columns used to fit the transformer. If `None`, all non-numeric columns are used.
    partial_order:
        The partial order of the labels. The labels are encoded in the order of the given list. Additional values are
        assigned labels in the order they are encountered during fitting.
    """

    # ------------------------------------------------------------------------------------------------------------------
    # Dunder methods
    # ------------------------------------------------------------------------------------------------------------------

    def __init__(
        self,
        *,
        column_names: str | list[str] | None = None,
        partial_order: list[Any] | None = None,
    ) -> None:
        super().__init__(column_names)

        if partial_order is None:
            partial_order = []

        # Parameters
        self._partial_order = partial_order

        # Internal state
        self._mapping: dict[str, dict[Any, int]] | None = None  # Column name -> value -> label
        self._inverse_mapping: dict[str, dict[int, Any]] | None = None  # Column name -> label -> value

    def __hash__(self) -> int:
        return _structural_hash(
            super().__hash__(),
            self._partial_order,
            # Leave out the internal state for faster hashing
        )

    # ------------------------------------------------------------------------------------------------------------------
    # Properties
    # ------------------------------------------------------------------------------------------------------------------

    @property
    def is_fitted(self) -> bool:
        """Whether the transformer is fitted."""
        return self._mapping is not None and self._inverse_mapping is not None

    @property
    def partial_order(self) -> list[Any]:
        """The partial order of the labels."""
        return list(self._partial_order)  # defensive copy

    # ------------------------------------------------------------------------------------------------------------------
    # Learning and transformation
    # ------------------------------------------------------------------------------------------------------------------

    def fit(self, table: Table) -> LabelEncoder:
        """
        Learn a transformation for a set of columns in a table.

        This transformer is not modified.

        Parameters
        ----------
        table:
            The table used to fit the transformer.

        Returns
        -------
        fitted_transformer:
            The fitted transformer.

        Raises
        ------
        ColumnNotFoundError
            If column_names contain a column name that is missing in the table.
        ValueError
            If the table contains 0 rows.
        """
        if self._column_names is None:
            column_names = [name for name in table.column_names if not table.get_column_type(name).is_numeric]
        else:
            column_names = self._column_names
            _check_columns_exist(table, column_names)
            _warn_if_columns_are_numeric(table, column_names)

        if table.row_count == 0:
            raise ValueError("The LabelEncoder cannot be fitted because the table contains 0 rows")

        # Learn the transformation
        mapping = {}
        reverse_mapping = {}

        for name in column_names:
            # Remember partial order
            mapping[name] = {value: index for index, value in enumerate(self._partial_order)}
            reverse_mapping[name] = {index: value for value, index in mapping[name].items()}

            unique_values = table.get_column(name).get_distinct_values()
            for value in unique_values:
                if value not in mapping[name]:
                    label = len(mapping[name])
                    mapping[name][value] = label
                    reverse_mapping[name][label] = value

        # Create a copy with the learned transformation
        result = LabelEncoder(column_names=column_names, partial_order=self._partial_order)
        result._mapping = mapping
        result._inverse_mapping = reverse_mapping

        return result

    def transform(self, table: Table) -> Table:
        """
        Apply the learned transformation to a table.

        The table is not modified.

        Parameters
        ----------
        table:
            The table to which the learned transformation is applied.

        Returns
        -------
        transformed_table:
            The transformed table.

        Raises
        ------
        TransformerNotFittedError
            If the transformer has not been fitted yet.
        ColumnNotFoundError
            If the input table does not contain all columns used to fit the transformer.
        ValueError
            If the table contains 0 rows.
        """
        import polars as pl

        # Used in favor of is_fitted, so the type checker is happy
        if self._column_names is None or self._mapping is None:
            raise TransformerNotFittedError

        _check_columns_exist(table, self._column_names)

        columns = [
            pl.col(name).replace(self._mapping[name], default=None, return_dtype=pl.UInt32)
            for name in self._column_names
        ]

        return Table._from_polars_lazy_frame(
            table._lazy_frame.with_columns(columns),
        )

    def inverse_transform(self, transformed_table: Table) -> Table:
        """
        Undo the learned transformation.

        The table is not modified.

        Parameters
        ----------
        transformed_table:
            The table to be transformed back to the original version.

        Returns
        -------
        original_table:
            The original table.

        Raises
        ------
        TransformerNotFittedError
            If the transformer has not been fitted yet.
        ColumnNotFoundError
            If the input table does not contain all columns used to fit the transformer.
        ColumnTypeError
            If the specified columns of the input table contain non-numerical data.
        """
        import polars as pl

        # Used in favor of is_fitted, so the type checker is happy
        if self._column_names is None or self._inverse_mapping is None:
            raise TransformerNotFittedError

        _check_columns_exist(transformed_table, self._column_names)
        _check_columns_are_numeric(
            transformed_table,
            self._column_names,
            operation="inverse-transform with a LabelEncoder",
        )

        columns = [pl.col(name).replace(self._inverse_mapping[name], default=None) for name in self._column_names]

        return Table._from_polars_lazy_frame(
            transformed_table._lazy_frame.with_columns(columns),
        )

is_fitted: bool

Whether the transformer is fitted.

partial_order: list[Any]

The partial order of the labels.

fit

Learn a transformation for a set of columns in a table.

This transformer is not modified.

Parameters:

Name Type Description Default
table Table

The table used to fit the transformer.

 required

Returns:

Name Type Description
fitted_transformer LabelEncoder

The fitted transformer.

Raises:

Type Description
ColumnNotFoundError

If column_names contain a column name that is missing in the table.
ValueError

If the table contains 0 rows.
Source code in src/safeds/data/tabular/transformation/_label_encoder.py 
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
def fit(self, table: Table) -> LabelEncoder:
    """
    Learn a transformation for a set of columns in a table.

    This transformer is not modified.

    Parameters
    ----------
    table:
        The table used to fit the transformer.

    Returns
    -------
    fitted_transformer:
        The fitted transformer.

    Raises
    ------
    ColumnNotFoundError
        If column_names contain a column name that is missing in the table.
    ValueError
        If the table contains 0 rows.
    """
    if self._column_names is None:
        column_names = [name for name in table.column_names if not table.get_column_type(name).is_numeric]
    else:
        column_names = self._column_names
        _check_columns_exist(table, column_names)
        _warn_if_columns_are_numeric(table, column_names)

    if table.row_count == 0:
        raise ValueError("The LabelEncoder cannot be fitted because the table contains 0 rows")

    # Learn the transformation
    mapping = {}
    reverse_mapping = {}

    for name in column_names:
        # Remember partial order
        mapping[name] = {value: index for index, value in enumerate(self._partial_order)}
        reverse_mapping[name] = {index: value for value, index in mapping[name].items()}

        unique_values = table.get_column(name).get_distinct_values()
        for value in unique_values:
            if value not in mapping[name]:
                label = len(mapping[name])
                mapping[name][value] = label
                reverse_mapping[name][label] = value

    # Create a copy with the learned transformation
    result = LabelEncoder(column_names=column_names, partial_order=self._partial_order)
    result._mapping = mapping
    result._inverse_mapping = reverse_mapping

    return result

fit_and_transform

Learn a transformation for a set of columns in a table and apply the learned transformation to the same table.

Note: Neither this transformer nor the given table are modified.

Parameters:

Name Type Description Default
table Table

The table used to fit the transformer. The transformer is then applied to this table.

 required

Returns:

Name Type Description
fitted_transformer Self

The fitted transformer.
transformed_table Table

The transformed table.
Source code in src/safeds/data/tabular/transformation/_table_transformer.py 
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
def fit_and_transform(self, table: Table) -> tuple[Self, Table]:
    """
    Learn a transformation for a set of columns in a table and apply the learned transformation to the same table.

    **Note:** Neither this transformer nor the given table are modified.

    Parameters
    ----------
    table:
        The table used to fit the transformer. The transformer is then applied to this table.

    Returns
    -------
    fitted_transformer:
        The fitted transformer.
    transformed_table:
        The transformed table.
    """
    fitted_transformer = self.fit(table)
    transformed_table = fitted_transformer.transform(table)
    return fitted_transformer, transformed_table

inverse_transform

Undo the learned transformation.

The table is not modified.

Parameters:

Name Type Description Default
transformed_table Table

The table to be transformed back to the original version.

 required

Returns:

Name Type Description
original_table Table

The original table.

Raises:

Type Description
TransformerNotFittedError

If the transformer has not been fitted yet.
ColumnNotFoundError

If the input table does not contain all columns used to fit the transformer.
ColumnTypeError

If the specified columns of the input table contain non-numerical data.
Source code in src/safeds/data/tabular/transformation/_label_encoder.py 
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
def inverse_transform(self, transformed_table: Table) -> Table:
    """
    Undo the learned transformation.

    The table is not modified.

    Parameters
    ----------
    transformed_table:
        The table to be transformed back to the original version.

    Returns
    -------
    original_table:
        The original table.

    Raises
    ------
    TransformerNotFittedError
        If the transformer has not been fitted yet.
    ColumnNotFoundError
        If the input table does not contain all columns used to fit the transformer.
    ColumnTypeError
        If the specified columns of the input table contain non-numerical data.
    """
    import polars as pl

    # Used in favor of is_fitted, so the type checker is happy
    if self._column_names is None or self._inverse_mapping is None:
        raise TransformerNotFittedError

    _check_columns_exist(transformed_table, self._column_names)
    _check_columns_are_numeric(
        transformed_table,
        self._column_names,
        operation="inverse-transform with a LabelEncoder",
    )

    columns = [pl.col(name).replace(self._inverse_mapping[name], default=None) for name in self._column_names]

    return Table._from_polars_lazy_frame(
        transformed_table._lazy_frame.with_columns(columns),
    )

transform

Apply the learned transformation to a table.

The table is not modified.

Parameters:

Name Type Description Default
table Table

The table to which the learned transformation is applied.

 required

Returns:

Name Type Description
transformed_table Table

The transformed table.

Raises:

Type Description
TransformerNotFittedError

If the transformer has not been fitted yet.
ColumnNotFoundError

If the input table does not contain all columns used to fit the transformer.
ValueError

If the table contains 0 rows.
Source code in src/safeds/data/tabular/transformation/_label_encoder.py 
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
def transform(self, table: Table) -> Table:
    """
    Apply the learned transformation to a table.

    The table is not modified.

    Parameters
    ----------
    table:
        The table to which the learned transformation is applied.

    Returns
    -------
    transformed_table:
        The transformed table.

    Raises
    ------
    TransformerNotFittedError
        If the transformer has not been fitted yet.
    ColumnNotFoundError
        If the input table does not contain all columns used to fit the transformer.
    ValueError
        If the table contains 0 rows.
    """
    import polars as pl

    # Used in favor of is_fitted, so the type checker is happy
    if self._column_names is None or self._mapping is None:
        raise TransformerNotFittedError

    _check_columns_exist(table, self._column_names)

    columns = [
        pl.col(name).replace(self._mapping[name], default=None, return_dtype=pl.UInt32)
        for name in self._column_names
    ]

    return Table._from_polars_lazy_frame(
        table._lazy_frame.with_columns(columns),
    )