资源简介

Rodriguez A, Laio A. Clustering by fast search and find of density peaks[J]. Science, 2014, 344(6191): 1492-1496.基于这篇文章实现的最基本的密度聚类的算法,具体请看我博客中的相关文章http://blog.csdn.net/kryolith/article/details/39832573

资源截图

代码片段和文件信息

import numpy as np
import matplotlib.pyplot as plt

from sklearn.datasets import make_blobs


def distanceNorm(NormD_value):
# initialization


# Norm for distance
if Norm == ‘1‘:
counter = np.absolute(D_value);
counter = np.sum(counter);
elif Norm == ‘2‘:
counter = np.power(D_value2);
counter = np.sum(counter);
counter = np.sqrt(counter);
elif Norm == ‘Infinity‘:
counter = np.absolute(D_value);
counter = np.max(counter);
else:
raise Exception(‘We will program this later......‘);

return counter;


def chi(x):
if x < 0:
return 1;
else:
return 0;


def fit(featureslabelstdistanceMethod = ‘2‘):
# initialization
distance = np.zeros((len(labels)len(labels)));
distance_sort = list();
density = np.zeros(len(labels));
distance_higherDensity = np.zeros(len(labels));


# compute distance
for index_i in xrange(len(labels)):
for index_j in xrange(index_i+1len(labels)):
D_value = features[index_i] - features[index_j];
distance[index_iindex_j] = distanceNorm(distanceMethodD_value);
distance_sort.append(distance[index_iindex_j]);
distance += distance.T;

# compute optimal cutoff
distance_sort = np.array(distance_sort);
cutoff = int(np.round(distance_sort[len(distance_sort) * t]));

# computer density
for index_i in xrange(len(labels)):
distance_cutoff_i = distance[index_i] - cutoff;
for index_j in xrange(1len(labels)):
density[index_i] += chi(distance_cutoff_i[index_j]);

# search for the max density
Max = np.max(density);
MaxIndexList = list();
for index_i in xrange(len(labels)):
if density[index_i] == Max:
MaxIndexList.extend([index_i]);

# computer distance_higherDensity
Min = 0;
for index_i in xrange(len(labels)):
if index_i in MaxIndexList:
distance_higherDensity[index_i] = np.max(distance[index_i]);
continue;
else:
Min = np.max(distance[index_i]);
for index_j in xrange(1len(labels)):
if density[index_i] < density[index_j] and distance[index_iindex_j] < Min:
Min = distance[index_iindex_j];
else:
continue;
distance_high

评论

共有 条评论