聚类 DBI 度量指标的数学公式与 Python 源码

数学公式见：

https://www.orzzz.net/directory/mathematics/MultivariateStatisticalAnalysis/CrispClusterValidityIndices/index.html

Python 源码：

def MinkowskiDistance(a: list, b: list, p: int = 2) -> float:
    return sum([abs(i - j) ** p for i, j in zip(a, b)]) ** (1 / p)


def Centroid(C: list[list]) -> list:
    return list(map(lambda c: sum(c) / len(c), zip(*C)))


def Cohesion(C: list[list]) -> float:
    centroid = Centroid(C)
    return sum(MinkowskiDistance(c, centroid) for c in C) / len(C)


def DaviesBouldinIndex(C: list[list[list]], DLL_SO: str = None) -> float:
    return sum(
        max(
            (Cohesion(Ci) + Cohesion(Cj)) / MinkowskiDistance(Centroid(Ci), Centroid(Cj)) for Cj in C if Ci != Cj
        )
        for Ci in C
    ) / len(C)


if __name__ == '__main__':
    C = [
        [
            [5.1, 3.5, 1.4, 0.2],
            [4.9, 3.0, 1.4, 0.2],
            [5.0, 3.3, 1.4, 0.2]
        ],

        [
            [4.9, 2.4, 3.3, 1.0],
            [6.6, 2.9, 4.6, 1.3]
        ],

        [
            [6.5, 3.0, 5.5, 1.8],
            [7.7, 3.8, 6.7, 2.2],
            [7.2, 3.2, 6.0, 1.8],
            [6.4, 2.8, 5.6, 2.1]
        ]
    ]
    print('Illusionna DBI:\t%.16f' % DaviesBouldinIndex(C))
    # --------------------------------------------------------------
    from sklearn.metrics import davies_bouldin_score
    matrix = [sample for cluster in C for sample in cluster]
    labels = [i for i in range(len(C)) for _ in range(len(C[i]))]
    print('sklearn DBI:\t%.16f' % davies_bouldin_score(matrix, labels))

与 sklearn 对比结果：