morphoclass.transforms.scalers package

Submodules

• morphoclass.transforms.scalers.abstract_scaler module
• morphoclass.transforms.scalers.manual_scaler module
• morphoclass.transforms.scalers.min_max_scaler module
• morphoclass.transforms.scalers.robust_scaler module
• morphoclass.transforms.scalers.standard_scaler module

Module contents

Various feature scalers for morphology node features.

class morphoclass.transforms.scalers.AbstractFeatureScaler(feature_indices, is_global_feature=False)

Bases: abc.ABC

Base abstract class for all feature scalers.

All derived classes must implement the _fit and _transform methods

Parameters

• feature_indices – List of indices of the feature maps to which to apply the scaling.

• is_global_feature – Apply scaler to global features rather than node features.

fit(dataset, idx=None)

Fit the scaler to data.

Parameters

• dataset (morphoclass.data.MorphologyDataset) – Data to which to fit the scaler.

• idx – Selects a subset of samples in the dataset to which to fit the data.

abstract get_config()

Generate the configuration necessary for reconstructing the scaler.

Returns

config – The configuration of the scaler. It should contain all information necessary for reconstructing the scaler using the scaler_from_config function.

Return type

dict

abstract reconstruct(params)

Reconstruct the configuration from parameters.

Parameters

params (dict) – The parameters found in config[“params”] with the config being the dictionary returned by get_config.

class morphoclass.transforms.scalers.FeatureManualScaler(feature_indices, shift=0, scale=1, **kwargs)

Bases: morphoclass.transforms.scalers.abstract_scaler.AbstractFeatureScaler

Scaler that shifts and scales the features by fixed values.

The new features are computed by features -> (features - shift) / scale

Parameters

• feature_indices – List of indices of the feature maps to which to apply the scaling.

• shift (float) – The fixed offset subtracted from the features

• scale (float) – The fixed scale by which the shifted features are divided.

• kwargs – Additional keyword argument to pass through to the AbstractFeatureScaler base class.

get_config()

Generate the configuration necessary for reconstructing the scaler.

Returns

config – The configuration of the scaler. It should contain all information necessary for reconstructing the scaler using the scaler_from_config function.

Return type

dict

reconstruct(params)

Reconstruct the configuration from parameters.

Parameters

params (dict) – The parameters found in config[“params”] with the config being the dictionary returned by get_config.

class morphoclass.transforms.scalers.FeatureMinMaxScaler(feature_indices, feature_range=(0, 1), take_abs=False, **kwargs)

Bases: morphoclass.transforms.scalers.abstract_scaler.AbstractFeatureScaler

Scaler that scales the features to a given range.

Internally the MinMaxScaler from scikit-learn is applied

Parameters

• feature_indices – List of indices of the feature maps to which to apply the scaling.

• feature_range (sequence (optional)) – The feature range to which to scale the features. This value is passed through to the MinMaxScaler class in sklearn.

• take_abs (bool (optional)) – If true then the scaler will be fitted on the absolute values of the features. This way a feature range of (0, 1) would translate to an effective range of (-1, 1). This is useful when it’s necessary to avoid shifting the features away from the origin.

• kwargs – Additional keyword argument to pass through to the AbstractFeatureScaler base class.

get_config()

Generate the configuration necessary for reconstructing the scaler.

Returns

config – The configuration of the scaler. It should contain all information necessary for reconstructing the scaler using the scaler_from_config function.

Return type

dict

reconstruct(params)

Reconstruct the configuration from parameters.

Parameters

params (dict) – The parameters found in config[“params”] with the config being the dictionary returned by get_config.

class morphoclass.transforms.scalers.FeatureRobustScaler(feature_indices, with_centering=True, with_scaling=True, **kwargs)

Bases: morphoclass.transforms.scalers.abstract_scaler.AbstractFeatureScaler

Scaler that is robust against outliers in data.

Internally the RobustScaler from scikit-learn is applied.

Parameters

• feature_indices – List of indices of the feature maps to which to apply the scaling.

• with_centering (bool (optional)) – If True, center the data before scaling. This value is passed through to the RobustScaler class in sklearn.

• with_scaling (bool (optional)) – If True, scale the data to interquartile range. This value is passed through to the RobustScaler class in sklearn.

• kwargs – Additional keyword argument to pass through to the AbstractFeatureScaler base class.

get_config()

Generate the configuration necessary for reconstructing the scaler.

Returns

config – The configuration of the scaler. It should contain all information necessary for reconstructing the scaler using the scaler_from_config function.

Return type

dict

reconstruct(params)

Reconstruct the configuration from parameters.

Parameters

params (dict) – The parameters found in config[“params”] with the config being the dictionary returned by get_config.

class morphoclass.transforms.scalers.FeatureStandardScaler(feature_indices, with_mean=True, with_std=True, **kwargs)

Bases: morphoclass.transforms.scalers.abstract_scaler.AbstractFeatureScaler

Scaler that removes the mean and standard deviation.

Internally the StandardScaler from scikit-learn is applied

Parameters

• feature_indices – List of indices of the feature maps to which to apply the scaling.

• with_mean (bool (optional)) – Whether or not to shift by the mean value. This value is passed through to the StandardScaler class in sklearn.

• with_std (bool (optional)) – Whether or not to scale by the standard deviation. This value is passed through to the StandardScaler class in sklearn.

• kwargs – Additional keyword argument to pass through to the AbstractFeatureScaler base class.

get_config()

Generate the configuration necessary for reconstructing the scaler.

Returns

config – The configuration of the scaler. It should contain all information necessary for reconstructing the scaler using the scaler_from_config function.

Return type

dict

reconstruct(params)

Reconstruct the configuration from parameters.

Parameters

params (dict) – The parameters found in config[“params”] with the config being the dictionary returned by get_config.

morphoclass.transforms.scalers.scaler_from_config(config)

Reconstruct scaler from a config.

Parameters

config (dict) – The configuration returned by get_config of a scaler class.

Returns

obj – The reconstructed scaler.

Return type

object