Transformer

Transformer

Transformer(wrapped_transformer[, columns])

Wrap any transformer with a .fit() and .transform() method for use with Pyreal.

Feature Select Transformer

FeatureSelectTransformer(columns, **kwargs)	A transformer that selects and re-orders features to match the model's inputs
ColumnDropTransformer(columns, **kwargs)	A transformer that drops a set of columns from the data

Scalers

MinMaxScaler([feature_range, clip])	Scales numeric features within a given range
StandardScaler(*[, with_mean, with_std])	Standardizes numeric features to mean=0 and variance=1
Normalizer([norm])	Normalizes numeric features using the l1, l2, or max norm

Imputers

MultiTypeImputer([columns])

Imputes a data set, handling columns of different types.

One-Hot Encoders

OneHotEncoder([columns, handle_unknown])	One-hot encodes categorical feature values
MappingsOneHotEncoder(mappings, **kwargs)	Converts data from categorical form to one-hot-encoded, with feature names based on a mappings object which includes two dictionaries
MappingsOneHotDecoder(mappings, **kwargs)	Converts data from one-hot encoded form to categorical, with feature names based on a mappings object which includes two dictionaries
Mappings.generate_mappings([one_to_many, ...])	Generate a new Mappings object using one of the input formats All but one keyword should be None

Time-Series Formatters

MultiIndexFrameToNestedFrame([model, ...])	Convert Pyreal time-series format to sktime nested DataFrame.
MultiIndexFrameToNumpy2d([model, interpret, ...])	This Transformer should only be used on univariate series Convert Pyreal time-series format into NumPy ndarray with shape (n_instances, n_timepoints).
MultiIndexFrameToNumpy3d([model, interpret, ...])	Convert Pyreal time-series format into NumPy ndarray with shape (n_instances, n_columns, n_timepoints).
NestedFrameToMultiIndexFrame([model, ...])	Convert sktime nested DataFrame format into Pyreal time-series format.
NestedFrameToNumpy3d([model, interpret, ...])	Convert sktime nested DataFrame format into NumPy ndarray with shape (n_instances, n_variables, n_timepoints).
Numpy2dToMultiIndexFrame([var_name, timestamps])	This Transformer should only be used on univariate time series data Convert 2D NumPy array to Pyreal time-series format.
Numpy2dToNestedFrame([var_name, timestamps])	Convert 2D NumPy array to sktime nested DataFrame.
Numpy3dToMultiIndexFrame([var_names, timestamps])	Convert 3D NumPy array to Pyreal time-series format.
Numpy3dToNestedFrame([var_names, timestamps])	Convert 3D NumPy array to sktime nested DataFrame.
Pandas2dToMultiIndexFrame([var_name, timestamps])	Convert 2D DataFrame to Pyreal time-series format.

Type Casters

BoolToIntCaster([model, interpret, ...])

Wrappers

DataFrameWrapper(wrapped_transformer, **kwargs)

Allows use of standard sklearn transformers while maintaining DataFrame type.

Time Series Padders

TimeSeriesPadder(value[, length])

A transformer that pads and truncates variable-length time series to equal lengths

Geo Transformers

LatLongToPlace([level, latitude_column, ...])

Converts latitude and longitude columns to neighborhood, city, or state/country names.

Narrative Transformers

NarrativeTransformer([llm, openai_api_key, ...])

Transforms explanations to narrative (natural-language) form.

Aggregating Transformers

Aggregator(mappings[, func, drop_original, ...])

Aggregate features into a single parent feature class