MCPcopy Create free account
hub / github.com/DynamicTimeWarping/dtw-python / StepPattern

Class StepPattern

dtw/stepPattern.py:28–377  ·  view source on GitHub ↗

Step patterns for DTW A ``stepPattern`` object lists the transitions allowed while searching for the minimum-distance path. DTW variants are implemented by passing one of the objects described in this page to the ``stepPattern`` argument of the [dtw()] call. **Details** A step pattern characteriz

Source from the content-addressed store, hash-verified

26
27
28class StepPattern:
29 # IMPORT_RDOCSTRING stepPattern
30 """Step patterns for DTW
31
32A ``stepPattern`` object lists the transitions allowed while searching
33for the minimum-distance path. DTW variants are implemented by passing
34one of the objects described in this page to the ``stepPattern``
35argument of the [dtw()] call.
36
37**Details**
38
39A step pattern characterizes the matching model and slope constraint
40specific of a DTW variant. They also known as local- or
41slope-constraints, transition types, production or recursion rules
42(GiorginoJSS).
43
44**Pre-defined step patterns**
45
46::
47
48 ## Well-known step patterns
49 symmetric1
50 symmetric2
51 asymmetric
52
53 ## Step patterns classified according to Rabiner-Juang (Rabiner1993)
54 rabinerJuangStepPattern(type,slope_weighting="d",smoothed=False)
55
56 ## Slope-constrained step patterns from Sakoe-Chiba (Sakoe1978)
57 symmetricP0; asymmetricP0
58 symmetricP05; asymmetricP05
59 symmetricP1; asymmetricP1
60 symmetricP2; asymmetricP2
61
62 ## Step patterns classified according to Rabiner-Myers (Myers1980)
63 typeIa; typeIb; typeIc; typeId;
64 typeIas; typeIbs; typeIcs; typeIds; # smoothed
65 typeIIa; typeIIb; typeIIc; typeIId;
66 typeIIIc; typeIVc;
67
68 ## Miscellaneous
69 mori2006;
70 rigid;
71
72A variety of classification schemes have been proposed for step
73patterns, including Sakoe-Chiba (Sakoe1978); Rabiner-Juang
74(Rabiner1993); and Rabiner-Myers (Myers1980). The ``dtw`` package
75implements all of the transition types found in those papers, with the
76exception of Itakura’s and Velichko-Zagoruyko’s steps, which require
77subtly different algorithms (this may be rectified in the future).
78Itakura recursion is almost, but not quite, equivalent to ``typeIIIc``.
79
80For convenience, we shall review pre-defined step patterns grouped by
81classification. Note that the same pattern may be listed under different
82names. Refer to paper (GiorginoJSS) for full details.
83
84**1. Well-known step patterns**
85

Callers 4

mvmStepPatternFunction · 0.90
TMethod · 0.85
rabinerJuangStepPatternFunction · 0.85
stepPattern.pyFile · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected