def init_with_kmeans(self): print(self.cols*self.rows) print(len(list(np.where(self._mask == 0))[1])) '''Initialise the BGDGMM and FGDGMM, which are respectively background-model and foreground-model, using kmeans algorithm''' max_iter = 2 # Max-iteration count for Kmeans '''In the following two indexings, the np.logical_or is needed in place of or''' self._bgd = np.where(np.logical_or(self._mask == self._GC_BGD, self._mask == self._GC_PR_BGD)) # Find the places where pixels in the mask MAY belong to BGD. self._fgd = np.where(np.logical_or(self._mask == self._GC_FGD, self._mask == self._GC_PR_FGD)) # Find the places where pixels in the mask MAY belong to FGD. self._BGDpixels = self.img[self._bgd] self._FGDpixels = self.img[self._fgd] KMB = kmeans(self._BGDpixels, dim = 3, n = self.k, max_iter = max_iter) # The Background Model by kmeans KMF = kmeans(self._FGDpixels, dim = 3, n = self.k, max_iter = max_iter) # The Foreground Model by kmeans KMB.run() KMF.run() # self._BGD_types = KMB.output() # self._FGD_types = KMF.output() # print(self._BGD_types) self._BGD_by_components = KMB.output() self._FGD_by_components = KMF.output() self.BGD_GMM = GMM() # The GMM Model for BGD self.FGD_GMM = GMM() # The GMM Model for FGD '''Add the pixels by components to GMM''' for ci in range(self.k): # print(len(self._BGD_by_components[ci])) # print(self._BGD_by_components[ci]) for pixel in self._BGD_by_components[ci]: # pixel = np.asarray([j for j in pixel], dtype = np.float32) self.BGD_GMM.add_pixel(pixel, ci) for pixel in self._FGD_by_components[ci]: self.FGD_GMM.add_pixel(pixel, ci) # for ci in range(self.k): # bgd_index = np.where(self._BGD_types == ci) # fgd_index = np.where(self._FGD_types == ci) # for pixel in self.img[bgd_index]: # self.BGD_GMM.add_pixel(pixel, ci) # for pixel in self.img[fgd_index]: # self.FGD_GMM.add_pixel(pixel, ci) self.BGD_GMM.learning() self.FGD_GMM.learning()
n = 100 * (index - 2)
n = index
print "K-means for " + str(n) + " centroids."
center = None
label = None
ret = 99999999999
for _ in range(1):
centroids = features[np.random.choice(range(len(features)),
n,
replace=False)]
c_ret, c_label, c_center = kmeans(features, k = n, \
centroids = centroids, steps = 100)
if c_ret < ret:
ret = c_ret
label = c_label
center = c_center
print ret
rets.append(ret)
save_clustering("features/hog_cluster.bin", c_label, c_center)
# plt.plot(rets)
# plt.show()
pid_list, command_list, param_list = [], [], []
index_0, index_1 = [], []
print('node: ', nodes[i][0])
for param in params[i]:
if param[2] is not None:
pid_list.append(param[0])
command_list.append(param[1])
param_list.append(param[2])
dataset = np.vstack((param_list, param_list))
dataset = np.transpose(dataset)
# kmeans clustering
k = 2
dataset = np.mat(dataset)
if len(dataset) > 0:
centroids, cluster_assment = km.kmeans(dataset, k)
for index, value in enumerate(cluster_assment[:, 0]):
if value == 0:
index_0.append(index)
if value == 1:
index_1.append(index)
print('kmeans: ', param_i)
if len(index_0) <= len(index_1):
print('command: ', [command_list[j] for j in index_0])
#print('pid: ', [pid_list[j] for j in index_0])
else:
print('command: ', [command_list[j] for j in index_1])
#print('pid: ', [pid_list[j] for j in index_0])
print()
# 2d plot
def k_means_RBFS(training_samples, N):
rbfs = kmeans(training_samples, N)
sigma = 0.1*( rbfs.max() - rbfs.min() )/ np.sqrt(2*N)
return rbfs, sigma
# coding=utf-8
import sys
import os
root_path = os.getcwd() # 获得当前py的根目录
sys.path.append(root_path + '/Other') # 导入Other文件夹
from OnePassCluster import *
import k_means
from sklearn import preprocessing
if __name__ == '__main__':
vectors = np.loadtxt('Other/dataSet/dim1024.txt') # k=16 threshold=0.1927
# vectors = np.loadtxt('Other/dataSet/g2-256-100.txt') # 0.192717136469125
# 归一化（Normalization）
vectors = preprocessing.normalize(vectors) # 归一化
# print vectors
k = 16
t1 = time.time()
cluster_result = np.array(k_means.kmeans(k=k, vectors=vectors))
t2 = time.time()
print "k-means spend time %.9fs" % ((t2 - t1) / 1000)
print cluster_result
for i in range(k):
print i, np.where(cluster_result == i)[0]
print '-------'
o_p_c = OnePassCluster(t=0.1927, vector_list=vectors)
o_p_c.print_result()
def _parse_address(self):
'''
Convert address to binary arrays.
'''
data = []
print "Parsing address."
print len(np.unique(self.data[:, 6]))
for index in range(len(self.data[:, 6])):
splitted_address = self.data[:, 6][index].split(' ', 1)
if self._is_int(splitted_address[0]):
data.append(splitted_address[1])
else:
data.append(self.data[:, 6][index])
self.data[index][6] = data[index]
labels = np.unique(data)
data = np.array(data)
means_address = []
for index_label in range(len(labels)):
address_data = self.data[np.where(data == labels[index_label])]
means_address.append([np.mean(address_data[:, 7]), np.mean(address_data[:, 8])])
means_address = np.float32(np.array(means_address))
cond_1 = means_address[np.where(means_address[:, 1] > 85)]
centroid1 = cond_1[0]
cond_2 = means_address[np.where(np.logical_and(means_address[:, 1] > 68, means_address[:, 1] < 80))]
centroid2 = cond_2[0]
cond_3 = means_address[np.where(np.logical_and(means_address[:, 1] > 50, means_address[:, 1] < 70))]
centroid3 = cond_3[0]
total = len(means_address)
# centroid1 = means_address[np.random.randint(total)]
# centroid2 = means_address[np.random.randint(total)]
# centroid3 = means_address[np.random.randint(total)]
rets = []
# for n in range(10, 11):
for n in range(100, 101):
center = None
label = None
ret = 9999999
for _ in range(5):
centroids = []
if n == 2:
centroids.append(centroid2)
centroids.append(means_address[np.random.randint(total)])
centroids = np.float32(centroids)
elif n == 3:
centroids.append(centroid1)
centroids.append(centroid2)
centroids.append(means_address[np.random.randint(total)])
centroids = np.float32(centroids)
elif n == 4:
centroids.append(centroid1)
centroids.append(centroid2)
centroids.append(centroid3)
centroids.append(means_address[np.random.randint(total)])
else:
centroids.append(centroid1)
centroids.append(centroid2)
centroids.append(centroid3)
centroids = np.float32(centroids)
centroids = np.vstack((centroids, means_address[np.random.choice(range(len(means_address)), n - 3)]))
while len(np.unique(centroids)) < n:
centroids = np.unique(centroids)
x = np.random.rand() * np.mean(centroids[:, 0])
y = np.random.rand() * np.mean(centroids[:, 1])
centroids = np.vstack((centroids, [x, y]))
c_ret, c_label, c_center = kmeans(means_address, k = n, \
centroids = centroids, steps = 1000)
if c_ret < ret:
ret = c_ret
label = c_label
center = c_center
print ret
rets.append(ret)
# plt.plot(rets)
# plt.show()
# Now separate the data, Note the flatten()
# Plot the data
# plt.scatter(means_address[:, 0], means_address[:, 1])
#
# plt.scatter(center[:, 0], center[:, 1], s = 80, c = 'y', marker = 's')
# plt.xlabel('Height'), plt.ylabel('Weight')
# plt.show()
for index_label in range(len(labels)):
data[np.where(data == labels[index_label])] = "A" + str(label[index_label])
labels = np.unique(data)
data = self._binarize_feature(data, labels)
del means_address
del splitted_address
del address_data
del rets
del center
del label
del c_label
del c_center
del centroids
return labels, data
from numpy import genfromtxt
from sklearn import svm, metrics
from sklearn.neighbors import KNeighborsClassifier
import scipy.io
from k_means import kmeans
data = scipy.io.loadmat('dataku.mat')["dataimage_plus"]
#print(data)
#data = genfromtxt('dataimage.csv', delimiter=',')
n_samples = len(data)
#split data input and data output
data_i = data[:,0:2] #2:4 => fft (mean dan standard deviasi FFT), 0:2 => (entropy dan energi GLCM)
data_o,c = kmeans(data_i,3)
print len(data_o)
data_expected = data[:,4] - 1
print len(data_expected)
#split 50 % data (data training)
# data_i_train = data_i[0:][::2]
# data_o_train = data_o[0:][::2]
# #split 50 % data (data testing)
# data_i_test = data_i[1:][::2]
# data_o_test = data_o[1:][::2]
# #create knn classifier
# neigh = KNeighborsClassifier(n_neighbors=3)
raw_data = mat73.loadmat('./data/kmeans_pts.mat')
gr_list = [t[0] for t in raw_data['Data']['gr_pts']]
n_itrs = 20
#finish this and run by tomorrow morning.
cluster_purities = pd.DataFrame(
columns=['Algorithm', 'NumberClusters', 'ClusterPurity'])
for k in range(5, 30, 5):
print('.')
print('.')
print('.')
for exp in range(10):
# for exp in range(2):
centers = km.kmeans(gr_list, k, n_itrs, 'flag')
cluster_purity = km.clusterPurity(labels_true, gr_list, centers,
'flag')
cluster_purities = cluster_purities.append(
{
'Algorithm': 'Flag Mean',
'NumberClusters': k,
'ClusterPurity': cluster_purity
},
ignore_index=True)
print("Flag trial" + str(exp + 1) + " finished")
print('.')
centers = km.kmeans(gr_list, k, n_itrs, 'sine')
cluster_purity = km.clusterPurity(labels_true, gr_list, centers,
'sine')
#pl.hist(useful_values,50, normed=1, facecolor='green', alpha=0.75)
#pl.show()
time_mat_PE1 = deepcopy(time_mat)
time_mat_PE2 = deepcopy(time_mat)
time_mat_PE3 = deepcopy(time_mat)
dim_max = np.shape(time_mat)
for i1 in range(0, dim_max[0]):
for j1 in range(0, dim_max[1]):
temp = double(time_mat[i1, j1]/500)
if temp > 1:
time_mat_PE2[i1, j1] = 500
if temp <0:
time_mat_PE2[i1, j1] = 0
centroids, clusterAssment = kmeans(time_mat, 5)
print (clusterAssment)
print (np.shape(clusterAssment))
#showCluster(time_mat, 5, centroids, clusterAssment)
######################################## POINT #####
first_line = ['pointId', 'lon', 'lat', 'alt', 'valueOfTime1', 'valueOfTime2', 'valueOfTime3', 'valueOfTime4', 'valueOfTime5', \
'valueOfTime6', 'valueOfTime7', 'valueOfTime8', 'valueOfTime9', 'valueOfTime10', 'valueOfTime11', 'valueOfTime12', 'valueOfTime13']
outfp_o_cl1.writerow(first_line)
outfp_o_cl2.writerow(first_line)
outfp_o_cl3.writerow(first_line)
outfp_o_cl4.writerow(first_line)
outfp_o_cl5.writerow(first_line)
print("=================================================================")
'''
K-Means Algorithm
'''
print("\n\nK-means Clustering")
k = input("Enter K: ")
max_iterations = input("Enter Maximum iterations: ")
min_df = input("Enter Minimum document Frequency: ")
print("=================================================================")
k = int(k)
max_iterations = int(max_iterations)
min_df = int(min_df)
start = time.time()
Kmeans = k_means.kmeans()
# Min Document Frequency is used as Feature Selcection Parameter
y_pred, labels = Kmeans.clustering(X, k, max_iterations, min_df)
# Vectors and Features of X in K-means Clustering
'''
k_means_vector = Kmeans.vectors
k_means_features = Kmeans.features
'''
contingency_matrix = metrics.cluster.contingency_matrix(y, y_pred)
score = purity_score(y, y_pred)
print("Purity: " + str(score))
end = time.time()
tot = end - start
from numpy import genfromtxt
from sklearn import svm, metrics
from sklearn.neighbors import KNeighborsClassifier
import scipy.io
from k_means import kmeans
data = scipy.io.loadmat('dataku.mat')["dataimage_plus"]
#print(data)
#data = genfromtxt('dataimage.csv', delimiter=',')
n_samples = len(data)
#split data input and data output
data_i = data[:,0:2] #2:4 => fft (mean dan standard deviasi FFT), 0:2 => (entropy dan energi GLCM)
data_o,c = kmeans(data_i,3)
#print data_o
data_exp = data[:,4]
#split 50 % data (data training)
data_i_train = data_i[0:][::2]
data_o_train = data_o[0:][::2]
#split 50 % data (data testing)
data_i_test = data_i[1:][::2]
data_o_test = data_exp[1:][::2] - 1
#create knn classifier
neigh = KNeighborsClassifier(n_neighbors=3)
#We learn the digits on the first half of the digits
# coding=utf-8 ''' Author: ripples Email: [email protected] date: 2020/3/11 15:34 desc: ''' # 可以做一个关于数据集是否混乱的对比 import sys import copy sys.path.append('../') import numpy as np import matplotlib.pyplot as plt from k_means import kmeans path = '../iris/iris.data' x = kmeans(path, 3) x.cal() # x.plot_label_true()
import scipy.io as sio
import numpy as np
import mat73
import center_algorithms as ca
import matplotlib.pyplot as plt
import k_means as km
import seaborn as sns
import pandas as pd
labels_raw = sio.loadmat(
'./data/kmeans_action_labels.mat')['kmeans_action_labels']
labels_true = [l[0][0] for l in labels_raw['labels'][0][0]]
# labelidxs =labels_raw['labelidxs'][0][0][0]
raw_data = mat73.loadmat('./data//kmeans_pts.mat')
gr_list = [t[0] for t in raw_data['Data']['gr_pts']]
n_itrs = 20
k = 15
centers = km.kmeans(gr_list, k, n_itrs, 'flag')
cluster_purity = km.clusterPurity(labels_true, gr_list, centers, 'flag')
import numpy as np import matplotlib.pyplot as plt from k_means import kmeans # исходные данные X = np.array([ [4, 4], [3, 3], [5, 3], [2, 3], [5, 5], [3, 2], [2, 4], [4, 5], [5, 4], [2, 2]]) # запуск кластеризации ans = kmeans(2, X) # отображение результатов print(ans) plt.plot(X[:,0], X[:,1], 'bx', ans[:,0], ans[:,1], 'r*', markersize=20) plt.grid() plt.show()
output = lda.final_output
with open("cluster_images/top10topicswith10wordswithoutprobs.csv",
"a") as top:
top.write("num_topics=" + str(num_topics) + "_no_above=" +
str(no_above).replace(".", "") + ",")
topics = lda.get_topics(num_words=10, probs=False)
for topic_words in topics:
top.write(str(topic_words) + ",")
top.write("\n")
with open("centroids.csv", "a") as top:
top.write(",")
for i in range(1, 11):
top.write("cluster" + str(i) + ",")
top.write("\n")
for k in [8]:
k_means = kmeans(output, "histograms")
labels, score = k_means.cluster(k)
with open("centroids.csv", "a") as top:
top.write("num_topics=" + str(num_topics) + "_no_above=" +
str(no_above).replace(".", "") + "_k=" + str(k) +
",")
centroids = k_means.centroids
for center in centroids:
for counter, topics in enumerate(center):
top.write(str(counter + 1) + "= " + str(topics) + " ")
top.write(",")
#top.write("\n")
k_means.plot_histogram2(
"num_topics=" + str(num_topics) + "_no_above=" +
"""
Created on Tue Oct 10 11:40:16 2017
@author: xuwh
"""
from numpy import *
import time
import matplotlib.pyplot as plt
import types
import k_means
## step 1: load data
print("step 1: load data...")
dataSet = []
fileIn = open(r'D:\Python\MachineLearningInAction\testSet.txt')
for line in fileIn.readlines():
lineArr = line.strip().split()
#print lineArr[0],lineArr[1]
#dataSet.append([float(lineArr[0]),float(lineArr[1])])
dataSet.append([float(lineArr[0])])
print("step 2: clustering...")
#这里使用mat将dataSet数据转换为矩阵之后才能进行线性代数操作
dataSet = mat(dataSet)
k = 4
centroids, clusterAssment = k_means.kmeans(dataSet, k)
# step 3: show the result
print("step 3: show the result...")
#plt.plot()
k_means.showCluster(dataSet, k, centroids, clusterAssment)
""" Created on Sat Oct 25 20:34:32 2014 @author: Imane """ import numpy as np import matplotlib.pyplot as plt #from os import listdir #from os.path import isfile, join #from zscoring import zscoringNII #from masking import maskdata from sklearn.decomposition import PCA from k_means import kmeans #Applying PCA and plotting fn = "dataZM\dataMask2.npy" d=np.load(fn) pca = PCA(n_components=2) pca.fit(d) dpca=pca.transform(d) plt.scatter(dpca[:,0], dpca[:,1], marker='o', color='b') #Applying kmeans and plotting idx, ctrs = kmeans(dpca, 2) plt.scatter(dpca[(idx==0),0], dpca[(idx==0),1], marker='o', color='r') plt.scatter(dpca[(idx==1),0], dpca[(idx==1),1], marker='o', color='b') plt.scatter(ctrs[:,0], ctrs[:,1], marker='o', color='k', linewidths=5)