Transformations

Transformations of the time series intended to be used in a similar fashion to torchvision. The user specifies a dictionary of Transformations in a particular order.

Example:

>>> transform = [
>>>     {'Tensorize': None},
>>>     {'LogTransform': {'targets': [0], 'offset': 1.0}},
>>>     {'RemoveLast': {'targets': [0]}},
>>>     {'Target': {'targets': [0]}}
>>> ]

class transforms.LogTransform(offset=0.0, targets=None)

Natural logarithm of target covariate + offset.

\[y_i = log_e ( x_i + \mbox{offset} )\]

class transforms.RemoveLast(targets=None)

Subtract last time series points from time series.

class transforms.Target(targets)

Retain only target covariates for output.

class transforms.Tensorize(device='cpu')

Convert ndarrays to Tensors.