Coverage for tdaad/utils/tda_functions.py: 100%

1"""Persistence Diagram Transformers.""" 

2 

3# Author: Martin Royer 

4 

5import numpy as np 

6 

7from gudhi.sklearn.rips_persistence import RipsPersistence 

8 

9 

10def _numpy_data_to_similarity(X, filter_nan=True): 

11 r"""Transforms numpy matrix X into similarity matrix :math:`1-\mathbf{Corr}(X)`.""" 

12 target = 1 - np.corrcoef(X, rowvar=False) 

13 # this filters when a variable is constant -> nan on all rows 

14 nanrowcols = np.isnan(target).all(axis=0) if filter_nan else ~target.any(axis=0) 

15 return target[~nanrowcols][:, ~nanrowcols] 

16 

17 

18def transform_to_persistence_diagram(X, tda_max_dim=0): 

19 """Persistence Diagram Transformer for point cloud. 

20 

21 For a given point cloud, form a similarity matrix and apply a RipsPersistence procedure 

22 to produce topological descriptors in the form of persistence diagrams. 

23 

24 Read more in the :ref: `User Guide <persistence_diagrams>`. 

25 

26 Parameters: 

27 tda_max_dim : int, default=0 

28 The maximum dimension of the topological feature extraction. 

29 

30 Example 

31 ------- 

32 >>> n_timestamps = 100 

33 >>> n_sensors = 5 

34 >>> import numpy as np 

35 >>> np.corrcoef(X) 

36 >>> import pandas as pd 

37 >>> timestamps = pd.to_datetime('2024-01-01', utc=True) + pd.Timedelta(1, 'h') * np.arange(n_timestamps) 

38 >>> X = pd.DataFrame(np.random.random(size=(n_timestamps, n_sensors)), index=timestamps) 

39 >>> PersistenceDiagramTransformer().fit_transform(X.to_numpy()) 

40 """ 

41 sim_target = [_numpy_data_to_similarity(X)] 

42 rips_transformer = RipsPersistence( 

43 homology_dimensions=range(tda_max_dim + 1), 

44 input_type="lower distance matrix", 

45 ) 

46 rips_target = rips_transformer.transform(sim_target) 

47 return rips_target[0]