k临近法的实现：kd树

 1 # coding:utf-8
 2 import numpy as np
 3 import matplotlib.pyplot as plt
 4
 5 T = [[2, 3], [5, 4], [9, 6], [4, 7], [8, 1], [7, 2]]
 6 S=[7, 3]
 7
 8 class node:
 9     def __init__(self, point):
10         self.left = None
11         self.right = None
12         self.point = point
13         self.parent = None
14         pass
15
16     def set_left(self, left):
17         if left == None: pass
18         left.parent = self
19         self.left = left
20
21     def set_right(self, right):
22         if right == None: pass
23         right.parent = self
24         self.right = right
25
26 def median(lst):
27     m = len(lst) / 2
28     return lst[m], m
29
30 def build_kdtree(data, d):
31     data = sorted(data, key=lambda x: x[d])
32     p, m = median(data)
33     tree = node(p)
34     del data[m]
35     if m > 0: tree.set_left(build_kdtree(data[:m], not d))
36     if len(data) > 1: tree.set_right(build_kdtree(data[m:], not d))
37     return tree
38
39 def distance(a, b):
40     return ((a[0] - b[0]) ** 2 + (a[1] - b[1]) ** 2) ** 0.5
41
42 def search_kdtree(tree, target,best=[]):
43     if len(best)==0: best = [tree.point,distance(tree.point, target)]
44     if target[0] < tree.point[0]:
45         if tree.left != None:
46             return search_kdtree(tree.left, target, best)
47     else:
48         if tree.right != None:
49             return search_kdtree(tree.right, target, best)
50     def update_best(t, best):
51         if t == None: return
52         t = t.point
53         d = distance(t, target)
54         if d < best[1]:
55             best[1] = d
56             best[0] = t
57     while (tree.parent != None):
58         update_best(tree.parent.left, best)
59         update_best(tree.parent.right, best)
60         tree = tree.parent
61     return best[0]
62
63 def showT(tree,d):
64     plt.plot(tree.point[0],tree.point[1],‘ob‘)
65     if tree.parent==None:
66         plt.plot([tree.point[0],tree.point[0]],[0,10])
67     elif d:
68         if tree.point[0]<tree.parent.point[0]:
69             plt.plot([0,tree.parent.point[0]],[tree.point[1],tree.point[1]])
70         else:
71             plt.plot([tree.parent.point[0],10],[tree.point[1],tree.point[1]])
72     else:
73         if tree.point[1]<tree.parent.point[1]:
74             plt.plot([tree.point[0],tree.point[0]],[0,tree.parent.point[1]])
75         else:
76             plt.plot([tree.point[0],tree.point[0]],[tree.parent.point[1],10])
77     if tree.left != None:
78         showT(tree.left,not d)
79     if tree.right != None:
80         showT(tree.right,not d)
81
82 kd_tree = build_kdtree(T, 0)
83 showT(kd_tree,0)
84 plt.annotate(‘S‘,xy = (S[0],S[1]+0.2))
85 plt.plot(S[0],S[1],‘^r‘)
86 result=search_kdtree(kd_tree,S)
87 print result  #[7, 2]
88 plt.show()