Transforms longitudinal exposure features to add the corresponding product features.
This preprocessor transform an input list of n_cases numpy arrays or
csr_matrices of shape (n_intervals, n_features) so as to add columns
representing the product of combination of two features. It outputs a list
of n_cases numpy arrays or csr_matrices of shape (n_intervals,
n_features + comb(n_features, 2)).
Exposure can take two forms: - short repeated exposures: in that case, each column of the numpy arrays or csr matrices can contain multiple ones, each one representing an exposure for a particular time bucket. - infinite unique exposures: in that case, each column of the numpy arrays or csr matrices can only contain a single one, corresponding to the starting date of the exposure.
exposure_type : {‘infinite’, ‘short’}, default=’infinite’
Either ‘infinite’ for infinite unique exposures or ‘short’ for short repeated exposures.
n_jobs : int, default=-1
Number of tasks to run in parallel. If set to -1, the number of tasks is set to the number of cores.
mapper : dict
Map product features to column indexes of the resulting matrices.
Examples
>>> from pprint import pprint
>>> from scipy.sparse import csr_matrix
>>> from tick.preprocessing.longitudinal_features_product import LongitudinalFeaturesProduct
>>> infinite_exposures = [csr_matrix([[0, 1, 0],
... [0, 0, 0],
... [0, 0, 1]], dtype="float64"),
... csr_matrix([[1, 1, 0],
... [0, 0, 1],
... [0, 0, 0]], dtype="float64")
... ]
>>> lfp = LongitudinalFeaturesProduct(exposure_type="infinite")
>>> product_features, _, _ = lfp.fit_transform(features)
>>> # output comes as a list of sparse matrices or 2D numpy arrays
>>> product_features.__class__
<class 'list'>
>>> [x.toarray() for x in product_features]
[array([[0., 1., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0.],
[0., 0., 1., 0., 0., 1.]]), array([[1., 1., 0., 1., 0., 0.],
[0., 0., 1., 0., 1., 1.],
[0., 0., 0., 0., 0., 0.]])]