ASV benchmark for scikit-learn/scikit-learn#20323

distance_metric.py

import numpy as np
from sklearn.neighbors import DistanceMetric
from .common import Benchmark

class DistanceMetricBenchmark(Benchmark):

    param_names = ["n", "d"]
    params = ([100, 1000, 10_000], [5, 10, 100])

    def setup(self, n, d):
        self.rng = np.random.RandomState(0)
        self.X = self.rng.random_sample((n, d))
        self.Y = self.rng.random_sample((n, d))

        self.V = self.rng.random_sample((d, d))
        self.VI = np.dot(self.V, self.V.T)

class EuclideanDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("euclidean")

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)


class ManhattanDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("manhattan")

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)


class ChebyshevDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("chebyshev")

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)


class MinkowskiDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("minkowski", p=1.5)

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)


class WMinkowskiDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("wminkowski", p=1.5,
                                                w=self.rng.random_sample(d))

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)


class SEuclideanDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("seuclidean", V=self.V)

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)



class MahalanobisDistanceBenchmark(DistanceMetricBenchmark):
    def setup(self, n, d):
        super().setup(n, d)
        self.dist_metric = DistanceMetric.get_metric("mahalanobis", VI=self.VI)

    def time_pairwise(self, n, d):
        self.dist_metric.pairwise(self.X, self.Y)