MapperPipeline¶

class gtda.mapper.pipeline.MapperPipeline(steps, memory=None, verbose=False)[source]¶

Subclass of sklearn.pipeline.Pipeline to deal with pipelines generated by make_mapper_pipeline.

The set_params method is modified from the corresponding method in sklearn.pipeline.Pipeline to allow for simple access to the parameters involved in the definition of the Mapper algorithm, without the need to interface with the nested structure of the Pipeline objects generated by make_mapper_pipeline. The convenience method get_mapper_params shows which parameters can be set. See the Examples below.

Examples

>>> from sklearn.cluster import DBSCAN
>>> from sklearn.decomposition import PCA
>>> from gtda.mapper import make_mapper_pipeline, CubicalCover
>>> filter_func = PCA(n_components=2)
>>> cover = CubicalCover()
>>> clusterer = DBSCAN()
>>> pipe = make_mapper_pipeline(filter_func=filter_func,
...                             cover=cover,
...                             clusterer=clusterer)
>>> print(pipe.get_mapper_params()['clusterer__eps'])
0.5
>>> pipe.set_params(clusterer___eps=0.1)
>>> print(pipe.get_mapper_params()['clusterer__eps'])
0.1