import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn import datasets
from sklearn import metrics
from scipy.spatial.distance import pdist,squareform
from collections import OrderedDict
def getDistCut(distList,distPercent):
maxDist = max(distList)
return maxDist * distPercent / 100
def getRho(n,distMatrix,distCut):
rho = np.zeros(n,dtype=float)
for i in range(n-1):
for j in range(i+1,n):
if distMatrix[i,j] < distCut:
rho[i] += 1
rho[j] += 1
print("rho:",rho[:10])
return rho
# def getRho(n,distMatrix,distCut):
# ro = np.zeros(n)
# for i in range(n-1):
# for j in range(i+1,n):
# ro[i] = ro[i] + np.exp(-(distMatrix[i, j] / distCut) ** 2)
# ro[j] = ro[j] + np.exp(-(distMatrix[i, j] / distCut) ** 2)
# ro = np.array([item / max(ro) for item in ro])
# return ro
def getGammaLeader(X,n,rho,distMatrix):
Delta = np.zeros(n,dtype=float)
Leader = np.ones(n,dtype=int) * (-1)
OrdRhoIndex = np.flipud(np.argsort(rho))
maxdist = 0
for i in range(n):
if distMatrix[OrdRhoIndex[0],i] > maxdist:
maxdist = distMatrix[OrdRhoIndex[0],i]
Delta[OrdRhoIndex[0]] = maxdist
'''获取密度最大点以外样本的Delta和Leader'''
for i in range(1,n):
mindist = np.inf
minindex = -1
for j in range(i):
if distMatrix[OrdRhoIndex[i],OrdRhoIndex[j]] < mindist:
mindist = distMatrix[OrdRhoIndex[i],OrdRhoIndex[j]]
minindex = OrdRhoIndex[j]
Delta[OrdRhoIndex[i]] = mindist
Leader[OrdRhoIndex[i]] = minindex
Gamma = Delta * rho
OrdGammaIndex = np.flipud(np.argsort(Gamma))
print("Gamma",len(Gamma))
# print(Gamma)
print("OrdGammaIndex",len(OrdGammaIndex))
print("Leader",len(Leader))
print(Leader)
EE = X[OrdGammaIndex[:3]]
plt.scatter(X[:,0],X[:,1],c = y,marker='o')
plt.scatter(EE[:,0],EE[:,1],marker='*',c='k')
plt.show()
return Gamma,OrdGammaIndex,OrdRhoIndex,Leader
def getInformationBlock(n,OrdGammaIndex,OrdRhoIndex,Leader):
blockNum = 2
clusterIndex = np.ones(n,dtype=int) * (-1)
leftBlock = []
rightBlock = []
for j in range(blockNum): ####直接给聚类中心点类簇标记{0,1}
clusterIndex[OrdGammaIndex[j]] = j
for i in range(1,n):
if clusterIndex[OrdRhoIndex[i]] == -1:
clusterIndex[OrdRhoIndex[i]] = clusterIndex[Leader[OrdRhoIndex[i]]]
print("clusterIndex",set(clusterIndex))
if len(set(clusterIndex)) != blockNum:
print("密度峰值聚类环节出错了：类簇索引不是两个")
for i in range(n):
if clusterIndex[i] == 0:
leftBlock.append(i)
elif clusterIndex[i] == 1:
rightBlock.append(i)
else:
print("出错了")
return leftBlock,rightBlock ####List类型
def fit(X,y,budget,distPercent):
n = len(X)
distList = pdist(X,metric='cityblock')
distMatrix = squareform(distList)
distCut = getDistCut(distList,distPercent)
rho = getRho(n,distMatrix,distCut)
Gamma, OrdGammaIndex,OrdRhoIndex, Leader = getGammaLeader(X,n,rho,distMatrix)
A,B = getInformationBlock(n, OrdGammaIndex,OrdRhoIndex, Leader)
return A,B
if __name__ == "__main__":
X,y = datasets.make_blobs(n_samples=200, n_features=2, centers=3,cluster_std=[1.0,1.0,1.0],random_state=100)
budget = 10
distPercent = 2
A,B = fit(X,y,budget,distPercent)
print("A+B:",len(A)+len(B))
AA = X[A]
BB = X[B]
plt.scatter(AA[:,0],AA[:,1],marker='o')
plt.scatter(BB[:,0],BB[:,1],marker='*')
plt.show()
