Highlevel API¶

On top of the base API this library already implements some useful classes:

An ArraySource: Source for numpy arrays with slicing functionaliy. And a SoundFileSource: Source for sound files using pysoundfile.
Commonly known spectral, temporal and other Features (e.g. RootMeanSquare or SpectralFlatness etc.)
A DefaultDictSink: Sink receiving results in dictionary.

Source¶

class sigfeat.source.array.ArraySource(array, samplerate, name='', **parameters)[source]¶

Source class for iterable arrays.

Parameters:	array : ndarray Expects an iterable array with .shape tuple. samplerate : int name : str blocksize : int overlap : int
Attributes:	`metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Returns generator that yields blocks out of the array.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

generate(self)[source]¶: Returns generator that yields blocks out of the array.

class sigfeat.source.soundfile.SoundFileSource(sf=None, **parameters)[source]¶

Source generating data from SoundFiles.

The parameters are for the SoundFile.blocks() method used for the blocks method of this Source class.

Parameters:

sf : SoundFile instance or str
blocksize : int
overlap : int
frames : int: The number of frames to yield from soundfile. If frames < 1, the file is read until the end.
fill_value : scalar: The value last block will filled up, if it is shorter tha blocksize.
dtype : {‘float64’, ‘float32’, ‘int32’, ‘int16’}, optional: See soundfile.SoundFile.read().
always_2d : bool: Indicates wether all blocks are at least 2d numpy arrays.

Attributes:

metadata: Returns metadata.
parameters: Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Returns generator that yields blocks from the SoundFile.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

generate(self)[source]¶: Returns generator that yields blocks from the SoundFile.

Feature¶

class sigfeat.feature.Index(name=None, requirements=None, **parameters)[source]¶

Index of source.

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

class sigfeat.feature.SpectralFlux(name=None, requirements=None, **parameters)[source]¶

Flux of AbsRfft.

\[ \begin{align}\begin{aligned}D_m[k] = \frac{|X_m[k]|}{\max(|X_m[k]|)} - \frac{|X_{m-1}[k]|}{\max(|X_{m-1}[k]|)}\\SFX_m = \frac{1}{2}\sum_k D[k]|D[k]|\end{aligned}\end{align} \]

Parameters:	axis : int Axis along the flux will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, featuredata)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

on_start(self, source, featureset, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, featuredata)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralRolloff(name=None, requirements=None, **parameters)[source]¶

Rolloff from AbsRfft.

The spectral rolloff is the frequency where the kappa percentage of energy is below and the 1-kappa percentage of energy is above.

# TODO formula

Parameters:	kappa : scalar {0…1} Default 0.95
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

on_start(self, source, featureset, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralCentroid(name=None, requirements=None, **parameters)[source]¶

Centroid of AbsRfft.

\[SC = \frac{\sum_k f[k]|X[k]|}{ \frac{1}{K}\sum_k |X[k]|}\]

Parameters:	axis : int Axis along the centroid will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

centroid
get_class_parameters

on_start(self, source, featureset, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralSpread(name=None, requirements=None, **parameters)[source]¶

Spread of AbsRfft.

\[SSP_m = \frac{\sum_k (f[k] SC_m[k])^2 |X_m[k]|}{ \sum_k |X_m[k]|}\]

Parameters:	axis : int Axis along the centroid will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

centroid
get_class_parameters
spread

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralSkewness(name=None, requirements=None, **parameters)[source]¶

Skewness of AbsRfft.

\[SSK_m = \frac{\sum_k (f[k]-SC_m[k])^3 |X_m[k]|}{ \sqrt{SSP_m}^3\sum_k |X_m[k]|}\]

Parameters:	axis : int Axis along the centroid will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

centroid
get_class_parameters
skewness

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralKurtosis(name=None, requirements=None, **parameters)[source]¶

Kurtosis of AbsRfft.

\[SK_m = \frac{\sum_k (f[k]-SC_m[k])^4 |X_m[k]|}{ SSP_m^2 \sum_k |X_m[k]|}\]

Parameters:	axis : int Axis along the centroid will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

centroid
get_class_parameters
kurtosis

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralFlatness(name=None, requirements=None, **parameters)[source]¶

Flatness of AbsRfft.

\[SF_m = \frac{\left(\Pi_k |X_m[k]| \right)^{1/K}}{ \frac{1}{K}\sum_k |X_m[k]|}\]

Parameters:	axis : int Axis along the flatness will be calculated, default=0.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, featuredata)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, featuredata)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralCrestFactor(name=None, requirements=None, **parameters)[source]¶

Crest Factor of AbsRfft

\[SCF_m = \frac{\max{(|X_m|)}}{X_{m, RMS}}\]

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.SpectralSlope(name=None, requirements=None, **parameters)[source]¶

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, features, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

on_start(self, source, features, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.MFCC(name=None, requirements=None, **parameters)[source]¶

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.RootMeanSquare(name=None, requirements=None, **parameters)[source]¶

Root Mean Square (RMS) from Source data.

\[RMS_m = \sqrt{\frac{1}{N}\sum_n x_m[n]^2}\]

Parameters:	axis : int Axis along which the feature is computed.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.CrestFactor(name=None, requirements=None, **parameters)[source]¶

Crest Factor of Source data.

Depends on Peak and RootMeanSquare features.

\[CF_m = \frac{\max{(|x_m|)}}{x_{m, RMS}}\]

Parameters:	axis: int
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, resd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.ZeroCrossingRate(name=None, requirements=None, **parameters)[source]¶

Zero Crossings Rate of Source data.

\[ZCR_m = \frac{f_s}{N} \lfloor 0.5 + \frac{1}{2} \sum_n |\mathrm{sgn}(x_m[n+1])-\mathrm{sgn}(x_m[n])| \rfloor\]

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

on_start(self, source, featureset, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

class sigfeat.feature.Peak(name=None, requirements=None, **parameters)[source]¶

Peak of AbsSignal from Source data.

\[P_m = \max(|x_m|)\]

Parameters:	axis : int Axis along which the feature is computed.
Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

class sigfeat.feature.Delta(feature, **parameters)[source]¶

Returns a diffenetiated version of the given feature.

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, resultd)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

process(self, data, resultd)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

Extractor¶

This Module contains the central Extractor class.

The Extractor consumes features instances and is used to extract the Features from given sources into given Sink.

class sigfeat.extractor.Extractor(*features, autoinst=True)[source]¶

Feature Extractor

Parameters:	features : Feature instances Provide multple feature instances. If you provide a Feature required by other features, please place it before the others. autoinst* : bool Autoinstantiate required feature Classes if no instance exists in featureset. If False you will get errors with a hint which feature instance you need provide as well.

Methods

`extract`(self, source[, sink])	Extracts features from given source into given sink.
`get_features_parameters_and_metadata`(self[, …])	Returns dict with parameters and metadata from self.featureset.
`get_parameters_and_metadata`(obj)	Returns dict with parameters and metadata from given `obj`.
`reset`(self)	Resets the states of features.

extract(self, source, sink=None)[source]¶

Extracts features from given source into given sink.

Parameters:	source : Source instance sink : Sink instance
Returns:	res : Sink The sink with processed data and metadata.

get_features_parameters_and_metadata(self, hidden=False)[source]¶: Returns dict with parameters and metadata from self.featureset.

static get_parameters_and_metadata(obj)[source]¶: Returns dict with parameters and metadata from given obj.

reset(self)[source]¶

Resets the states of features.

If a new source shall be processed this may be usefull or needed.

Sink¶

class sigfeat.sink.default.DefaultDictSink[source]¶

The receive_append method appends input to ‘results’ defaultdict(list)

Methods

`clear`()
`copy`()
`fromkeys`(iterable[, value])	Create a new dictionary with keys from iterable and values set to value.
`get`(self, key[, default])	Return the value for key if key is in the dictionary, else default.
`items`()
`keys`()
`pop`()	If key is not found, d is returned if given, otherwise KeyError is raised
`popitem`()	2-tuple; but raise KeyError if D is empty.
`receive`(self, datad)	Updates the dict with given `datad` dictionary.
`receive_append`(self, resultd)	Appends given `resultd` dict to list fields in this dict.
`setdefault`(self, key[, default])	Insert key with a value of default if key is not in the dictionary.
`update`()	If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
`values`()

receive(self, datad)[source]¶: Updates the dict with given datad dictionary.

receive_append(self, resultd)[source]¶: Appends given resultd dict to list fields in this dict.

Preprocess¶

class sigfeat.preprocess.mix.MeanMix(source, **parameters)[source]¶

Averages multi channel source e.g stereo to mono mix.

Parameters:	axis : int channels : int Number of channels will be generated by this preprocess.
Attributes:	`metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Override this method.
`process`(self, data)	Override this method.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

process(self, data)[source]¶

Override this method.

Parameters:	data : data from source

class sigfeat.preprocess.mix.SumMix(source, **parameters)[source]¶

Sums up multi channel source.

Parameters:	axis : int channels : int Number of channels will be generated by this preprocess.
Attributes:	`metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Override this method.
`process`(self, data)	Override this method.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

process(self, data)[source]¶

Override this method.

Parameters:	data : data from source

Base level API¶

Contains the abstract base classes.

Source¶

Implements abstract Source base class

class sigfeat.base.source.Source(**parameters)[source]¶

Base Source Class.

The parameters are mandatory for all kinds of sources. Every source must implement a .generate() method returning a generator that yields the signal blocks. Source must have samplerate and channels as metadata.

Parameters:	blocksize : int overlap : int samplerate : scalar channels : int
Attributes:	`metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Override this method.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

generate(self)[source]¶

Override this method. It must yield data.

Usually data = (block, index)

Feature¶

This module implements the abstract base Feature class.

Feature is subclassed for implementing Signal Features.

TODO: What happens if i have a hidden feature but add a instance of tha same that is not hidden? Mus be solved in the FeatureSet! TODO: Handle labels for multidimensional feature output.

class sigfeat.base.feature.Feature(name=None, requirements=None, **parameters)[source]¶

Abstract Feature Base Class

Parameters:	name : str requirements : Feature instances iterable To override those returned by self.requires().

Notes

At least the process() method must be overridden. If your implemented feature depends on the results of another feature, you must override the requires() method returning an iterable e.g. list of feature instances.

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

dependencies(self)[source]¶: Yields all dependencies of this feature.

featureset(self, new=False, autoinst=False, err_missing=True)[source]¶

Returns an ordered dict of all features unique in name.

The dict is ordered by the dependency tree order.

Parameters:	new : boolean All features will be reinitialized.
Returns:	featdict : OrderedDict Keys are `fid` and values are feature instances.

fid¶: Returns the feature identifying tuple.

gen_dependencies_instances(self, autoinst=False, err_missing=True)[source]¶: Checks deps for being instance or class and yields instances.

hidden¶: Returns whether the feature is hidden or not.

hide(self, b=True)[source]¶: Hide the feature.

new(self)[source]¶: Returns new initial feature instance with same parameters.

on_finished(self, source, featureset, sink)[source]¶

Override this method to be run after extraction.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

on_start(self, source, featureset, sink)[source]¶

Override this method if your feature needs some initialization.

The Extractor will give you source, featureset and sink.

Parameters:	source : source featureset : OrderedDict of features sink : Sink

process(self, data, result)[source]¶

Override this method returning process results.

The data from the source will be in data (usually data = block,index,...,).

The required results from other features will be in the result argument (dict like Result()).

Parameters:	data : block, index Or the data generated from your source. result : Result The results from other features of current data. If you provide your required features in the requires() method, you will be able to access the results from those features.
Returns:	res : e.g. scalar, ndarray or other custom types. This is the feature result for the current data (block).

requires(self)[source]¶

Override this method if your feature depends on other features.

You can return Feature classes and Feature instances you need for your feature. Yielding is also allowed.

validate_name(self)[source]¶: Checks for uniqueness of feature name in all dependent features.

class sigfeat.base.feature.HiddenFeature(name=None, requirements=None, **parameters)[source]¶

Attributes:	`fid` Returns the feature identifying tuple. `hidden` Returns whether the feature is hidden or not. `metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`dependencies`(self)	Yields all dependencies of this feature.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`featureset`(self[, new, autoinst, err_missing])	Returns an ordered dict of all features unique in name.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`gen_dependencies_instances`(self[, autoinst, …])	Checks deps for being instance or class and yields instances.
`hide`(self[, b])	Hide the feature.
`new`(self)	Returns new initial feature instance with same parameters.
`on_finished`(self, source, featureset, sink)	Override this method to be run after extraction.
`on_start`(self, source, featureset, sink)	Override this method if your feature needs some initialization.
`process`(self, data, result)	Override this method returning process results.
`requires`(self)	Override this method if your feature depends on other features.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.
`validate_name`(self)	Checks for uniqueness of feature name in all dependent features.

get_class_parameters

sigfeat.base.feature.features_to_featureset(features, new=False, autoinst=False)[source]¶

Returns a featureset of given features distinct in names.

Parameters:	features : iterable new : reinitialize features as new instances. autoinst : auto initialize missing feature classes if required.

Result¶

class sigfeat.base.result.Result[source]¶

Result dict. Behaves ‘immutable’ to the Feature.process method.

Just a simple dict to hold the results from features.

Methods

clear()

copy()

fromkeys(iterable[, value]) Create a new dictionary with keys from iterable and values set to value.

get(self, key[, default]) Return the value for key if key is in the dictionary, else default.

items()

keys()

pop() If key is not found, d is returned if given, otherwise KeyError is raised

popitem() 2-tuple; but raise KeyError if D is empty.

setdefault(self, key[, default]) Insert key with a value of default if key is not in the dictionary.

update() If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]

values()

Sink¶

Implements the abstract Sink class.

class sigfeat.base.sink.Sink[source]¶

Sink base class.

Methods

`receive`(datad)	Shall receive dictionaries directly written to source.
`receive_append`(resultd)	Shall receive result dictionaries appending data to fields.

classmethod receive(datad)[source]¶: Shall receive dictionaries directly written to source.

classmethod receive_append(resultd)[source]¶: Shall receive result dictionaries appending data to fields.

Preprocess¶

Implements a simple preprocess base class.

For processing blocks of a source before extracting features. Preprocess behaves like a Source and is consumed by extractor or another Preprocess.

class sigfeat.base.preprocess.Preprocess(source, **parameters)[source]¶

Preprocess base class.

Behaves like a source. But you mus ovrride the process method.

Examples

>>> src = YourPreprocess(YourSource(...))
>>> extractor.extract(src, ...)

Attributes:	`metadata` Returns metadata. `parameters` Returns all parameters.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.
`generate`(self)	Override this method.
`process`(self, data)	Override this method.
`unroll_parameters`(self, parameters)	This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

generate(self)[source]¶

Override this method. It must yield data.

Usually data = (block, index)

process(self, data)[source]¶

Override this method.

Parameters:	data : data from source

Parameter¶

This module implements the Parameter and it’s mixin class.

Purpose is, to distinguish parameters of objects from other attributes. Parameters of instances will be extracted into Sink.

class sigfeat.base.parameter.Parameter(default=None)[source]¶

Adds a Parameter to the class.

Attributes:	`default` Returns the default value.

Methods

validate(self, value) You can override this method for validation of parameter values.

default¶: Returns the default value.

validate(self, value)[source]¶: You can override this method for validation of parameter values.

class sigfeat.base.parameter.ParameterMixin(**parameters)[source]¶

ParameterMixin class

Adds Parameter functionality to classes.

Attributes:	`parameters` Returns all parameters.

Methods

unroll_parameters(self, parameters) This method must be called to collect all parameters and provide them as attributes.

get_class_parameters

parameters¶: Returns all parameters.

unroll_parameters(self, parameters)[source]¶

This method must be called to collect all parameters and provide them as attributes.

By calling this function defined parameters will become attributes with values not beeing of type Parameter anymore. But all parameters will be placed in self._parameters as well.

Metadata¶

Mixin for Metadata.

Adds methods add_metadata(), extend_metadata() and fetch_metadata_as_attrs() to the class.

Metadata will be extracted into Sink. So created feature datasets contain metadata for e.g. features and sources.

class sigfeat.base.metadata.MetadataMixin[source]¶

MetadataMixin class Adding metadata functionality to classes. Overrides:

._metadata
.metadata
._init_metadata_list
.add_metadata
.extend_metadata
.fetch_metadata_as_attrs

Attributes:	`metadata` Returns metadata.

Methods

`add_metadata`(self, name, value)	Appends the key value pair to metadata list.
`extend_metadata`(self, mdata)	Extends the metadata with the given list of key value pairs.
`fetch_metadata_as_attrs`(self)	Sets metadata as attributes of self.

add_metadata(self, name, value)[source]¶: Appends the key value pair to metadata list.

extend_metadata(self, mdata)[source]¶: Extends the metadata with the given list of key value pairs.

fetch_metadata_as_attrs(self)[source]¶: Sets metadata as attributes of self.

metadata¶: Returns metadata.