import numpy as np
from numpy import *
import sys
from sklearn import metrics
import pandas as pd
import matplotlib.pyplot as plt
def fit_plot_Kmean_model(X):
data = pd.read_csv(X, header=None, error_bad_lines=False)
# 必须添加header=None，否则默认把第一行数据处理成列名导致缺失
data_list = data.values.tolist()
data_list = np.array(data_list)
return data_list
def choice_center(data,k):
k=int(k)
centers = []
for i in np.random.choice(len(data), k):
centers.append(data[i])
# print("随机选取的中心点(第一次):\n", centers)
return centers
def distance(a, b):
dis = []
for i in range(len(a)):
dis.append(pow(a[i] - b[i], 2))
return sqrt(sum(dis))
def k_center(data_list,center,n_clusters,savePath):
flag = True
i = 0
a=len(data_list[0])-1
n_clusters=int(n_clusters)
while flag:
flag = False
for i in range(len(data_list)): # 遍历所有样本，最后一列标记该样本所属簇
min_index = -2
min_dis = inf
for j in range(len(center)):
dis = distance(data_list[i][1:n_clusters],center[j][1:n_clusters])
if dis < min_dis:
min_dis = dis
min_index = j
if data_list[i][-1] != min_index:
flag = True
data_list[i][-1] = min_index
# print("分类结果111：",data_list)
# 重新计算簇中心
for k in range(len(center)): # 遍历中心向量，取出属于当前中心向量簇的样本
current_k = []
for i in range(len(data_list)):
if data_list[i][-1] == k:
current_k.append(data_list[i])
# print(k, "：", current_k)
old_dis = 0.0
for i in range(len(current_k)):
old_dis += distance(current_k[i][1:n_clusters], center[k][1:n_clusters])
for m in range(len(current_k)):
new_dis = 0.0
for n in range(len(current_k)):
new_dis += distance(current_k[m][1:n_clusters], current_k[n][1:n_clusters])
if new_dis < old_dis:
old_dis = new_dis
center[k][:] = current_k[m][:]
# print("选中的最终中心点", center)
for i in range(len(data_list)): # 遍历所有样本，最后一列标记该样本所属簇
min_index = -2
min_dis = inf
for j in range(len(center)):
dis = distance(data_list[i][1:n_clusters], center[j][1:n_clusters])
if dis < min_dis:
min_dis = dis
min_index = j
data_list[i][-1] = min_index
# print("分类结果222：", data_list)
y_pred=data_list[:,a]
# print("y_pred：", y_pred)
score=metrics.calinski_harabaz_score(data_list,y_pred)
plt.figure(figsize=(12, 12))
plt.xticks(())
plt.yticks(())
plt.subplot(221)
plt.scatter(data_list[:, 0], data_list[:, 1])
plt.title("Raw data")
plt.subplot(222)
plt.scatter(data_list[:, 0], data_list[:, 1], c=y_pred)
plt.title("Kmediods:k={},score={}".format(n_clusters, int(score)))
# plt.title("DBSCAN:k={},eps={},min_samples={}".format(n_clusters, eps, min_samples))
# 字典中的key值即为csv中列名
dataframe = pd.DataFrame(y_pred)
# 将DataFrame存储为csv,index表示是否显示行名，default=True
savePath = savePath + "\\Kmediods_result.csv"
dataframe.to_csv(savePath, header=False, index=False, sep=',')
plt.show()
a = 'OK'
return a
if __name__ == '__main__':
data_list=fit_plot_Kmean_model(sys.argv[2])
# 1为选取的中心点个数，2为文件打开路径,3为文件保存路径
centers = choice_center(data_list,sys.argv[1])
print(k_center(data_list, centers,sys.argv[1],sys.argv[3]))
# data_list = fit_plot_Kmean_model('C:\\Users\\upc\\Desktop\\testdata.csv')
# # 1为选取的中心点个数，2为文件打开路径,3为文件保存路径
# centers = choice_center(data_list, 4)
# print(k_center(data_list, centers,4, 'C:\\Users\\upc\\Desktop'))
# 1为选取的中心点个数，2为文件打开路径,3为文件保存路径
# print(k_center(fit_plot_Kmean_model(sys.argv[2]), choice_center(fit_plot_Kmean_model(sys.argv[2]),sys.argv[1]),sys.argv[1],sys.argv[3]))
# print(k_center(fit_plot_Kmean_model('C:\\Users\\upc\\Desktop\\testdata.csv'), choice_center(fit_plot_Kmean_model('C:\\Users\\upc\\Desktop\\testdata.csv'),4),4,'C:\\Users\\upc\\Desktop'))
hiddenimports=['pandas','cython','numpy', 'sklearn','sklearn.metrics','sklearn.metrics.get_scorer','sklearn.metrics.cluster','sklearn.ensemble','sklearn.neighbors.typedefs','sklearn.neighbors.quad_tree','sklearn.tree._utils','scipy._lib.messagestream','email.mime.message', 'email.mime.image', 'email.mime.multipart', 'email.mime.audio', 'email.mime.text'],
