gaitmap.data_transform: Scaler and Feature Transforms for IMU data

Classes representing data transformations as preprocessing for different algorithms.

Base classes

BaseTransformer()	Base class for all data transformers.
BaseFilter()	Base class for all filters.
TrainableTransformerMixin()	Mixin for transformers with adaptable parameters.

Higher-level Transformers

GroupedTransformer(transformer_mapping[, ...])	Apply specific transformations to specific groups of columns.
ChainedTransformer(chain)	Apply a series of transformations to the input.
ParallelTransformer(transformers)	Apply multiple different transformation to the input, resulting in multiple outputs.

Simple Transformers and Scalers

IdentityTransformer()	Dummy Transformer that does not modify the data and simply returns a copy of the input.
FixedScaler([scale, offset])	Apply a fixed scaling and offset to the data.
StandardScaler([ddof])	Apply a standard scaling to the data.
TrainableStandardScaler([mean, std, ddof])	Apply a standard scaling to the data.
AbsMaxScaler([out_max])	Scale data by its absolute maximum.
TrainableAbsMaxScaler([out_max, data_max])	Scale data by the absolut max of a trainings sequence.
MinMaxScaler([out_range])	Scale the data by its Min-Max values.
TrainableMinMaxScaler([out_range, data_range])	Scale the data by Min-Max values learned from trainings data.

Feature Transformers

Resample([target_sampling_rate_hz])	Resample a time series using the scipy resample method.
BaseSlidingWindowFeatureTransform([...])	Baseclass for all Sliding window feature transforms.
SlidingWindowMean([window_size_s])	Calculate a sliding window mean.
SlidingWindowVar([window_size_s])	Calculate a sliding window variance.
SlidingWindowStd([window_size_s])	Calculate a sliding window standard deviation.
SlidingWindowGradient([window_size_s])	Calculate a sliding gradient by fitting a linear function on every sliding window.

Filter

ButterworthFilter(order, cutoff_freq_hz[, ...])

Apply a forward-backward (filtfilt) butterworth filter using the transformer interface.