scipy.spatial.KDTree.query中distance_upper_bound参数的度量标准是什么？ [英] What is the metric for distance_upper_bound parameter in scipy.spatial.KDTree.query?

问题描述

我不熟悉scipy。我似乎不太清楚在 scipy.spatial.KDTree.query 中 distance_upper_bound 的指标。是公里还是弧度？

I am new to using scipy. I can't quite seem to figure out the metric of distance_upper_bound in scipy.spatial.KDTree.query. Is it in Kilometers or Radians?

推荐答案

这就是值根据您决定使用参数 p 的度量标准。

It's exactly the value in terms of the chosen metric on which you decided using parameter p.

如果确实选择了 p = 1 ，又称曼哈顿距离，向量的距离： x = [1,2,3] 和 y = [2,3,4] 是 3 。如果您使用 distance_upper_bound = 2 并且 y 是 x的下一个邻居您在寻找，不要期望正确的结果。

If you did chose p=1, aka manhattan-distance, the distance of the vectors: x=[1,2,3] and y=[2,3,4] is 3. If you would have used distance_upper_bound=2 and y is the next neighbor to x you are looking for, don't expect a correct result.

备注：您正在谈论的此参数设置为 inf 默认情况。

Remark: this parameter you are talking about is set to inf by default.

您的任务似乎与纬度/经度有关。在这种情况下，我认为您想使用 Haversine-metric （免责声明：我不是这方面的专家）。

Your task seems to be about latitude/longitude points. In this case, i think you want to use the Haversine-metric (disclaimer: i'm no expert in this area).

遗憾的是，根据 p规范，这是scipy的邻居搜索中唯一支持的规范！

Sadly, this metric is imho not available in terms of a p-norm, the only ones supported in scipy's neighbor-searches!

但是：sklearn的 BallTree 可以与Haversine一起使用！ （KDTree不会！仍然是p范数！）

But: sklearn's BallTree can work with Haversine! (KDTree does not! Still p-norms!)

大概有一个很好的理由（无论是数学还是实际表现）为何KDTree不支持Haversine，而BallTree却支持。不要试图盲目地滚动自己的Haversine-KDTree！

There is probably a good reason (either math or practical performance) why KDTree is not supporting Haversine, while BallTree does. Don't try to roll your own Haversine-KDTree blindly!

来自：

From here:

二叉搜索树无法通过设计处理极坐标表示的环绕。您可能需要将坐标转换为3D笛卡尔空间，然后应用自己喜欢的搜索算法，例如kD-Tree，Octree等。

A binary search tree cannot handle the wraparound of the polar representation by design. You might need to transform the coordinates to a 3D cartesian space and then apply your favorite search algorithm, e.g., kD-Tree, Octree etc.

from sklearn.neighbors import KDTree, BallTree

KDTree.valid_metrics
# ['euclidean', 'l2', 'minkowski', 'p', 'manhattan', 'cityblock', 'l1', 'chebyshev',
#  'infinity']

BallTree.valid_metrics
# ['euclidean', 'l2', 'minkowski', 'p', 'manhattan', 'cityblock', 'l1', 'chebyshev',
#  'infinity', 'seuclidean', 'mahalanobis', 'wminkowski', 'hamming', 'canberra',
#  'braycurtis', 'matching', 'jaccard', 'dice', 'kulsinski', 'rogerstanimoto', 'russellrao',
#  'sokalmichener', 'sokalsneath', 'haversine', 'pyfunc']

这篇关于scipy.spatial.KDTree.query中distance_upper_bound参数的度量标准是什么？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！