synthesized3.meta package

class synthesized3.meta.Meta

Bases: ABC, BaseModel

Abstract base class & Pydantic model to store meta info about a tabular data column.

Parameters:

• name (str) – Name of the column the meta information being stored is about.

• nature (Nature) – Relates to whether the data is interpreted categorically or continuously

• observed_dtype (ColumnType) – The storage datatype of the column, converted to our internal column type ColumnType. E.g. data stored in a database as a VARCHAR(24) here would be saved as STRING, or similar.

• missing_value_meta (MissingValueMeta) – Optional instance of the MissingValueMeta submeta if it exists for this column.

name: str

observed_dtype: ColumnType

nature: Nature

missing_value_meta: Meta | None

get_sub_metas()

Recursively get the sub metas.

Parameters:

meta (Meta) – Meta to get sub metas from.

Returns:

List of sub metas.

Return type:

List[Meta]

class synthesized3.meta.MetaCollection

Bases: Mapping[str, Meta]

A class to store a collection of Meta objects. This class is iterable and is indexed by the names of columns in tables. Notably, the keys are specific to the original data_interface and not to any expanded or transformed data interfaces. Columns that are sub-columns of original data columns (e.g. _nan columns for columns that contain NaNs, or _dow, _month, … columns for datetime columns) are stored in the submeta field of the original column’s Meta object.

__init__(metas: Sequence[Meta], num_rows: int)

property metas: List[Meta]

property num_rows: int

classmethod from_data_interface(data_interface: DataInterface, meta_overrides: Mapping[str, Type[Meta]] | None = None, num_quantiles: PositiveInt = 100) → MetaCollection

Factory for creating MetaCollection objects

class synthesized3.meta.BooleanMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing booleans

categories: List[Any]

frequencies: List[float]

nature: Literal[Nature.CATEGORICAL]

missing_value_meta: MissingValueMeta | None

classmethod freqs_sum_to_one(values)

class synthesized3.meta.CategoricalMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing categories

categories: List[Any]

frequencies: List[float]

nature: Literal[Nature.CATEGORICAL]

missing_value_meta: MissingValueMeta | None

classmethod freqs_sum_to_one(values)

classmethod len_freqs_categs_equal(values)

property num_categories

class synthesized3.meta.ConstantMeta

Bases: Meta

Meta class to store information about columns which contain a single non-missing value

categories: List[Any]

nature: Literal[Nature.CATEGORICAL]

missing_value_meta: MissingValueMeta | None

class synthesized3.meta.DatetimeMeta

Bases: Meta

Metaclass to store information about columns which are to be treated as containing datetimes. For now, restricted to be continuous in nature. The unix_meta property can be an instance of DoubleMeta or LongMeta depending on the required accuracy of the unix timestamp

unix_meta: DoubleMeta

hour_meta: CategoricalMeta | ConstantMeta

dow_meta: CategoricalMeta | ConstantMeta

day_meta: CategoricalMeta | ConstantMeta

month_meta: CategoricalMeta | ConstantMeta

year_meta: CategoricalMeta | ConstantMeta

missing_value_meta: MissingValueMeta | None

nature: Literal[Nature.CONTINUOUS]

classmethod check_categories_are_within_allowed_limits(values)

get_sub_metas()

Recursively get the sub metas.

Parameters:

meta (Meta) – Meta to get sub metas from.

Returns:

List of sub metas.

Return type:

List[Meta]

class synthesized3.meta.DoubleMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing floats

quantiles: List[float]

missing_value_meta: MissingValueMeta | None

nature: Literal[Nature.CONTINUOUS]

class synthesized3.meta.FloatMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing floats

quantiles: List[float]

nature: Literal[Nature.CONTINUOUS]

missing_value_meta: MissingValueMeta | None

class synthesized3.meta.IntegerMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing integers

quantiles: List[float]

nature: Literal[Nature.CONTINUOUS]

missing_value_meta: MissingValueMeta | None

class synthesized3.meta.LongMeta

Bases: Meta

Meta class to store information about columns which are to be treated as containing integers

quantiles: List[float]

missing_value_meta: MissingValueMeta | None

nature: Literal[Nature.CONTINUOUS]

Subpackages

• synthesized3.meta.metas package
  • BooleanMeta
    • BooleanMeta.categories
    • BooleanMeta.frequencies
    • BooleanMeta.nature
    • BooleanMeta.missing_value_meta
    • BooleanMeta.freqs_sum_to_one()
    • BooleanMeta.name
    • BooleanMeta.observed_dtype
  • CategoricalMeta
    • CategoricalMeta.categories
    • CategoricalMeta.frequencies
    • CategoricalMeta.nature
    • CategoricalMeta.missing_value_meta
    • CategoricalMeta.freqs_sum_to_one()
    • CategoricalMeta.len_freqs_categs_equal()
    • CategoricalMeta.num_categories
    • CategoricalMeta.name
    • CategoricalMeta.observed_dtype
  • DatetimeMeta
    • DatetimeMeta.unix_meta
    • DatetimeMeta.hour_meta
    • DatetimeMeta.dow_meta
    • DatetimeMeta.day_meta
    • DatetimeMeta.month_meta
    • DatetimeMeta.year_meta
    • DatetimeMeta.missing_value_meta
    • DatetimeMeta.nature
    • DatetimeMeta.check_categories_are_within_allowed_limits()
    • DatetimeMeta.get_sub_metas()
    • DatetimeMeta.name
    • DatetimeMeta.observed_dtype
  • DoubleMeta
    • DoubleMeta.quantiles
    • DoubleMeta.missing_value_meta
    • DoubleMeta.nature
    • DoubleMeta.name
    • DoubleMeta.observed_dtype
  • FloatMeta
    • FloatMeta.quantiles
    • FloatMeta.nature
    • FloatMeta.missing_value_meta
    • FloatMeta.name
    • FloatMeta.observed_dtype
  • IntegerMeta
    • IntegerMeta.quantiles
    • IntegerMeta.nature
    • IntegerMeta.missing_value_meta
    • IntegerMeta.name
    • IntegerMeta.observed_dtype
  • LongMeta
    • LongMeta.quantiles
    • LongMeta.missing_value_meta
    • LongMeta.nature
    • LongMeta.name
    • LongMeta.observed_dtype
  • Submodules
  • synthesized3.meta.metas.boolean_meta module
    • BooleanMeta
      • BooleanMeta.categories
      • BooleanMeta.frequencies
      • BooleanMeta.nature
      • BooleanMeta.missing_value_meta
      • BooleanMeta.freqs_sum_to_one()
      • BooleanMeta.name
      • BooleanMeta.observed_dtype
  • synthesized3.meta.metas.boolean_meta_test module
  • synthesized3.meta.metas.categorical_meta module
    • CategoricalMeta
      • CategoricalMeta.categories
      • CategoricalMeta.frequencies
      • CategoricalMeta.nature
      • CategoricalMeta.missing_value_meta
      • CategoricalMeta.freqs_sum_to_one()
      • CategoricalMeta.len_freqs_categs_equal()
      • CategoricalMeta.num_categories
      • CategoricalMeta.name
      • CategoricalMeta.observed_dtype
  • synthesized3.meta.metas.categorical_meta_test module
  • synthesized3.meta.metas.constant_meta module
    • ConstantMeta
      • ConstantMeta.categories
      • ConstantMeta.nature
      • ConstantMeta.missing_value_meta
      • ConstantMeta.name
      • ConstantMeta.observed_dtype
  • synthesized3.meta.metas.constant_meta_test module
  • synthesized3.meta.metas.datetime_meta module
    • DatetimeMeta
      • DatetimeMeta.unix_meta
      • DatetimeMeta.hour_meta
      • DatetimeMeta.dow_meta
      • DatetimeMeta.day_meta
      • DatetimeMeta.month_meta
      • DatetimeMeta.year_meta
      • DatetimeMeta.missing_value_meta
      • DatetimeMeta.nature
      • DatetimeMeta.check_categories_are_within_allowed_limits()
      • DatetimeMeta.get_sub_metas()
      • DatetimeMeta.name
      • DatetimeMeta.observed_dtype
  • synthesized3.meta.metas.datetime_meta_test module
  • synthesized3.meta.metas.double_meta module
    • DoubleMeta
      • DoubleMeta.quantiles
      • DoubleMeta.missing_value_meta
      • DoubleMeta.nature
      • DoubleMeta.name
      • DoubleMeta.observed_dtype
  • synthesized3.meta.metas.double_meta_test module
  • synthesized3.meta.metas.float_meta module
    • FloatMeta
      • FloatMeta.quantiles
      • FloatMeta.nature
      • FloatMeta.missing_value_meta
      • FloatMeta.name
      • FloatMeta.observed_dtype
  • synthesized3.meta.metas.float_meta_test module
  • synthesized3.meta.metas.integer_meta module
    • IntegerMeta
      • IntegerMeta.quantiles
      • IntegerMeta.nature
      • IntegerMeta.missing_value_meta
      • IntegerMeta.name
      • IntegerMeta.observed_dtype
  • synthesized3.meta.metas.integer_meta_test module
  • synthesized3.meta.metas.long_meta module
    • LongMeta
      • LongMeta.quantiles
      • LongMeta.missing_value_meta
      • LongMeta.nature
      • LongMeta.name
      • LongMeta.observed_dtype
  • synthesized3.meta.metas.long_meta_test module
  • synthesized3.meta.metas.missing_value_meta module
  • synthesized3.meta.metas.missing_value_meta_test module

Submodules

synthesized3.meta.meta module

class synthesized3.meta.meta.Meta

Bases: ABC, BaseModel

Abstract base class & Pydantic model to store meta info about a tabular data column.

Parameters:

• name (str) – Name of the column the meta information being stored is about.

• nature (Nature) – Relates to whether the data is interpreted categorically or continuously

• observed_dtype (ColumnType) – The storage datatype of the column, converted to our internal column type ColumnType. E.g. data stored in a database as a VARCHAR(24) here would be saved as STRING, or similar.

• missing_value_meta (MissingValueMeta) – Optional instance of the MissingValueMeta submeta if it exists for this column.

name: str

observed_dtype: ColumnType

nature: Nature

missing_value_meta: Meta | None

get_sub_metas()

Recursively get the sub metas.

Parameters:

meta (Meta) – Meta to get sub metas from.

Returns:

List of sub metas.

Return type:

List[Meta]

synthesized3.meta.meta_collection module

class synthesized3.meta.meta_collection.MetaCollection

Bases: Mapping[str, Meta]

A class to store a collection of Meta objects. This class is iterable and is indexed by the names of columns in tables. Notably, the keys are specific to the original data_interface and not to any expanded or transformed data interfaces. Columns that are sub-columns of original data columns (e.g. _nan columns for columns that contain NaNs, or _dow, _month, … columns for datetime columns) are stored in the submeta field of the original column’s Meta object.

__init__(metas: Sequence[Meta], num_rows: int)

property metas: List[Meta]

property num_rows: int

classmethod from_data_interface(data_interface: DataInterface, meta_overrides: Mapping[str, Type[Meta]] | None = None, num_quantiles: PositiveInt = 100) → MetaCollection

Factory for creating MetaCollection objects

synthesized3.meta.meta_collection_test module

synthesized3.meta.meta_factory module

synthesized3.meta.meta_factory_test module

synthesized3.meta.meta_test module