MetricMethodClassification/MetricMethod/Metrics.py at main · DimaOshchepkov/MetricMethodClassification · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
from abc import ABC, abstractmethod

import numpy as np
import pandas as pd
from scipy.spatial.distance import cdist

__all__ = [
            'IMetric',
            'MetricsFactory',
            'ManhattanMetric',
            'EuclideanMetric',
            'CosineMetric']

class IMetric(ABC):

    __instance = None

    def __new__(cls, *args, **kwargs):
        if cls.__instance is None:
            cls.__instance = super().__new__(cls)

        return cls.__instance

    def __del__(self):
        IMetric.__instance = None


    @abstractmethod
    def get_distance(self, data : pd.Series, point : pd.Series) -> float:
        """
        Args:
            data (pd.Series): vector1
            point (pd.Series): vector2

        Returns:
            float: distance between data and point
        """
        pass

    @abstractmethod
    def get_distance_matrix(self, vectors : pd.DataFrame) -> np.ndarray:
        """
        Args:
            data (pd.DataFrame): vectors

        Returns
        -------
        np.ndarray: distance matrix
        """
        raise NotImplementedError


class ManhattanMetric(IMetric):

    def get_distance(self, data : pd.Series, point : pd.Series) -> float:
        return np.sum(np.abs(data - point), axis=-1)

    def get_distance_matrix(self, vectors : pd.DataFrame) -> np.ndarray:
        return cdist(vectors, vectors, metric='cityblock')


class EuclideanMetric(IMetric):

    def get_distance(self, data : pd.Series, point : pd.Series) -> float:
        return np.linalg.norm(data - point, axis=-1)

    def get_distance_matrix(self, vectors : pd.DataFrame) -> np.ndarray:
        return cdist(vectors, vectors, metric='euclidean')


class CosineMetric(IMetric):

    def get_distance(self, data : pd.Series, point : pd.Series) -> float:
        return (1 - data.dot(point) /
                    (np.linalg.norm(data, axis=-1) * np.linalg.norm(point, axis=-1)))

    def get_distance_matrix(self, vectors : pd.DataFrame) -> np.ndarray:
        return cdist(vectors, vectors, metric='cosine')


class MetricsFactory:
    """Use to create metrics"""

    __instance = None

    def __new__(cls, *args, **kwargs):
        if cls.__instance is None:
            cls.__instance = super().__new__(cls)

        return cls.__instance

    def __del__(self):
        MetricsFactory.__instance = None

    __exist_metrics : list = ["euclidean", "cityblock", 'cosine']

    @property
    def exist_metrics(self) -> list:
        return self.__exist_metrics

    def metrics_exist(self, name : str) -> bool:
        return name in self.__exist_metrics

    def get_metrics(self, name : str) -> IMetric:
        """
        Use to create metrics

        Args:
            name (str): name of metric
            Exists metrics: euclidean, cityblock, cosine

        Returns:
            IMetric: class of metric
        """
        if name == "euclidean":
            return EuclideanMetric()
        elif name == "cityblock":
            return ManhattanMetric()
        elif name == 'cosine':
            return CosineMetric()