Question

我正在实现K均值聚类算法。到目前为止，这就是我所拥有的：

import copy
import csv
import math
import random
import sys


class Centroid():
    def __init__(self, coordinates, _id):
      self.id = _id
      self.coordinates = coordinates
      self.elements = []

    def __repr__(self):
       return 'Centroid: ' + str(self.id)

    @property
    def count(self):
       return len(self.elements)



    def recalculate_coordinates(self):
       x = [sum(y)/len(y) for y in zip(*self.elements)]
       self.coordinates = x

    def reset_elements(self):
       self.previous_elements = []
       for el in self.elements:
          self.previous_elements.append(el)
       self.elements = []

class Kmeans():
  def __init__(self):
     self.k = int(sys.argv[2])
     self.prepare_data()
     self.iterations = 0

  def prepare_data(self):
     filename = sys.argv[1]
     self.dataset = []
     with open(filename, 'rb') as csvfile:
        reader = csv.reader(csvfile, delimiter=' ')
        for row in reader:
            tuplified = tuple(map(float, row))
            self.dataset.append(tuplified)
     self.create_centroids()

  def create_centroids(self):
    self.centroids = []
    for i in xrange(self.k):
        chosen = random.choice(self.dataset)
        cent = Centroid(chosen, i+1)
        self.centroids.append(cent)

def main():
  k = Kmeans()
  def iterate(k):
    k.iterations += 1
    for item in k.dataset:
        candidates = []
        for centroid in k.centroids:
            z = zip(item, centroid.coordinates)
            squares = map(lambda x: (x[0]-x[1])**2, z)  
            added = sum(squares)
            edistance = math.sqrt(added)
            candidates.append((centroid, edistance))
        winner = min(candidates, key=lambda x: x[1])
        winner[0].add_element(item)
    for centroid in k.centroids:
        centroid.reset_elements()
        centroid.recalculate_coordinates()

    status_list = []
    for centroid in k.centroids:
        boole = sorted(centroid.elements) == sorted(centroid.previous_elements)
        status_list.append(boole)

    if False in status_list:
        iterate(k)
    print k.centroids
    print k.iterations
  iterate(k)


if __name__ == '__main__':
   main()

但是，我一直收到错误RuntimeError: maximum recursion depth exceeded in cmp。我尝试了几个重构没有成功。谁能告诉我可能是什么问题。提前谢谢。

Answer 1

如果错误出现在这一行：

boole = sorted(centroid.elements) == sorted(centroid.previous_elements)

最有可能发生的是您在centroids.elements和centroids.previous_elements中有循环引用，因此比较操作（在sorted调用和==中执行）继续周期性地遍历每个列表。

这种行为的简单演示（在Python 3中）：

>>> x = []
>>> y = [x]
>>> x.append(y)
>>> x == y
Traceback (most recent call last)
  ....
   x == y    
RecursionError: maximum recursion depth exceeded in comparison

RuntimeError：cmp中超出的最大递归深度：K表示聚类

1 个答案: