|
1
|
|
|
# from abc import ABC, abstractmethod |
|
2
|
|
|
|
|
3
|
|
|
# from collections import Counter |
|
4
|
|
|
# import attr |
|
5
|
|
|
|
|
6
|
|
|
|
|
7
|
|
|
# class CountsComputerInterface(ABC): |
|
8
|
|
|
# @abstractmethod |
|
9
|
|
|
# def compute_counts(self, *args, **kwargs): |
|
10
|
|
|
# raise NotImplementedError |
|
11
|
|
|
|
|
12
|
|
|
# class AbstractCountsComputer(CountsComputerInterface, ABC): |
|
13
|
|
|
# def __call__(self, *args, **kwargs): |
|
14
|
|
|
# return self.compute_counts(*args, **kwargs) |
|
15
|
|
|
|
|
16
|
|
|
|
|
17
|
|
|
# @attr.s |
|
18
|
|
|
# class BaseCountsComputer(AbstractCountsComputer): |
|
19
|
|
|
|
|
20
|
|
|
# def compute_counts(self, *args, **kwargs): |
|
21
|
|
|
# return Counter(args[0]) |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
# # @attr.s |
|
25
|
|
|
# # class DatapointsCountsComputer(AbstractCountsComputer): |
|
26
|
|
|
# # counts_computer = attr.ib(init=True) |
|
27
|
|
|
# # |
|
28
|
|
|
# # def compute_counts(self, datapoints, attributes): |
|
29
|
|
|
# # return {attrib: self.counts_computer(self._iter(datapoints, attrib)) for attrib in attributes} |
|
30
|
|
|
# # |
|
31
|
|
|
# # def _iter(self, datapoints, k): |
|
32
|
|
|
# # return iter(self._extract(d, k) for d in datapoints) |
|
33
|
|
|
# # |
|
34
|
|
|
# # def _extract(self, datapoint, attribute): |
|
35
|
|
|
# # return getattr(datapoint, attribute) |
|
36
|
|
|
|
|
37
|
|
|
|
|
38
|
|
|
# @attr.s |
|
39
|
|
|
# class ClusterCountsComputer(AbstractCountsComputer): |
|
40
|
|
|
# datapoints_counts_computer = attr.ib(init=True) |
|
41
|
|
|
# extractor = attr.ib(init=True) |
|
42
|
|
|
|
|
43
|
|
|
# def compute_counts(self, cluster, attributes): |
|
44
|
|
|
# return self.datapoints_counts_computer(self.extractor(cluster), attributes) |
|
45
|
|
|
|
|
46
|
|
|
# ###################### |
|
47
|
|
|
|
|
48
|
|
|
# class DistroComputerInterface(ABC): |
|
49
|
|
|
# @abstractmethod |
|
50
|
|
|
# def compute_frequencies(self, *args, **kwargs): |
|
51
|
|
|
# raise NotImplementedError |
|
52
|
|
|
|
|
53
|
|
|
# class AbstractDistroComputer(DistroComputerInterface, ABC): |
|
54
|
|
|
# def __call__(self, *args, **kwargs): |
|
55
|
|
|
# return self.compute_frequencies(*args, **kwargs) |
|
56
|
|
|
|
|
57
|
|
|
|
|
58
|
|
|
# @attr.s |
|
59
|
|
|
# class BaseDistroComputer(AbstractDistroComputer): |
|
60
|
|
|
# counts_computer = attr.ib(init=True) |
|
61
|
|
|
# norm = attr.ib(init=False, default=None) |
|
62
|
|
|
|
|
63
|
|
|
# def compute_frequencies(self, *args, **kwargs): |
|
64
|
|
|
# _ = self.counts_computer(*args, **kwargs) |
|
65
|
|
|
# self.norm = sum(_.values()) |
|
66
|
|
|
# return _ / self.norm |
|
67
|
|
|
|
|
68
|
|
|
|
|
69
|
|
|
# @attr.s |
|
70
|
|
|
# class DatapointsDistroComputer(AbstractDistroComputer): |
|
71
|
|
|
# distro_computer = attr.ib(init=True) |
|
72
|
|
|
# extractor = attr.ib(init=True) |
|
73
|
|
|
|
|
74
|
|
|
# def compute_frequencies(self, datapoints, attributes): |
|
75
|
|
|
# return {attrib: self.distro_computer(self.extractor(datapoints, attrib)) for attrib in attributes} |
|
76
|
|
|
|
|
77
|
|
|
|
|
78
|
|
|
# @attr.s |
|
79
|
|
|
# class ClusterDistroComputer(AbstractDistroComputer): |
|
80
|
|
|
# datapoints_distro_computer = attr.ib(init=True) |
|
81
|
|
|
# extractor = attr.ib(init=True) |
|
82
|
|
|
|
|
83
|
|
|
# def compute_frequencies(self, cluster, attributes): |
|
84
|
|
|
# return self.datapoints_distro_computer(self.extractor(cluster), attributes) |
|
85
|
|
|
|
|
86
|
|
|
# @classmethod |
|
87
|
|
|
# def from_extractors(cls, extractor1, extractor2): |
|
88
|
|
|
# """Extractor 1 (cluster) -> datapoints, Extractor 2 (datapoints, attribute)""" |
|
89
|
|
|
# return ClusterDistroComputer(DatapointsDistroComputer(BaseDistroComputer(BaseCountsComputer()), extractor2), extractor1) |
|
90
|
|
|
|