#coding:utf-8 import math Iris_setosa_data=[] Iris_versicolor_data=[] Iris_virginica_data=[] #读取训练数据集，这里我将每种花取前45条数据，剩下的5条数据另外存入测试数据集 def read_train_data(filename): f=open(filename,'r') all_lines=f.readlines() for line in all_lines[0:45]: line=line.strip().split(',') Iris_setosa_data.append(line[0:4]) #Iris_setosa_label+=1 for line in all_lines[51:95]: line=line.strip().split(',') Iris_versicolor_data.append(line[0:4]) #Iris_versicolor_label+=1 for line in all_lines[101:145]: line=line.strip().split(',') Iris_virginica_data.append(line[0:4]) #Iris_virginica_label+=1 return Iris_setosa_data,Iris_versicolor_data,Iris_virginica_data test_data=[] #读取测试数据集 def read_test_data(testname): f=open(testname,'r') all_lines=f.readlines() for line in all_lines[0:]: line=line.strip().split(',') #以逗号为分割符拆分列表 test_data.append(line) return test_data #计算均值和方差 def calculate_junzhi_and_fangcha(train_data): x1_sum=0.0 x2_sum=0.0 x3_sum=0.0 x4_sum=0.0 for x in train_data: #计算各个特征的和 x1_sum+=float(x[0]) x2_sum+=float(x[1]) x3_sum+=float(x[2]) x4_sum+=float(x[3]) #print(x[0],x[1],x[2],x[3]) #计算样本在各个属性上取值的均值 u_x1=x1_sum/45 u_x2=x2_sum/45 u_x3=x3_sum/45 u_x4=x4_sum/45 k1=0.0 k2=0.0 k3=0.0 k4=0.0 #计算各类样本在第i个属性上的方差 for x in train_data: k1+=(float(x[0])-u_x1)**2 k2+=(float(x[1])-u_x2)**2 k3+=(float(x[2])-u_x3)**2 k4+=(float(x[3])-u_x4)**2 variance_x1=k1/45 variance_x2=k2/45 variance_x3=k3/45 variance_x4=k4/45 return u_x1,u_x2,u_x3,u_x4,variance_x1,variance_x2,variance_x3,variance_x4 #计算每个属性估计条件概率 def calculate_P_xi_c(u_x1,u_x2,u_x3,u_x4,variance_x1,variance_x2,variance_x3,variance_x4,line_data): p_x1_c=(1/math.sqrt(2*math.pi))*math.exp(-(float(line_data[0])-u_x1)**2/(2*variance_x1)) p_x2_c=(1/math.sqrt(2*math.pi))*math.exp(-(float(line_data[1])-u_x2)**2/(2*variance_x2)) p_x3_c=(1/math.sqrt(2*math.pi))*math.exp(-(float(line_data[2])-u_x3)**2/(2*variance_x3)) p_x4_c=(1/math.sqrt(2*math.pi))*math.exp(-(float(line_data[3])-u_x4)**2/(2*variance_x4)) return p_x1_c,p_x2_c,p_x3_c,p_x4_c if __name__ == '__main__': filename='iris_data.txt' testname='iris_test_data.txt' Iris_setosa_data,Iris_versicolor_data,Iris_virginica_data=read_train_data(filename) #Iris_setosa种类的各个特征属性上的均值和方差 Iris_setosa_u_x1,Iris_setosa_u_x2,Iris_setosa_u_x3,Iris_setosa_u_x4,\ Iris_setosa_variance_x1,Iris_setosa_variance_x2,Iris_setosa_variance_x3,\ Iris_setosa_variance_x4=calculate_junzhi_and_fangcha(Iris_setosa_data) #Iris_versicolor种类的各个特征属性上的均值和方差 Iris_versicolor_u_x1,Iris_versicolor_u_x2,Iris_versicolor_u_x3,Iris_versicolor_u_x4,\ Iris_versicolor_variance_x1,Iris_versicolor_variance_x2,Iris_versicolor_variance_x3,\ Iris_versicolor_variance_x4=calculate_junzhi_and_fangcha(Iris_versicolor_data) #Iris_virginica种类的各个特征属性上的均值和方差 Iris_virginica_u_x1,Iris_virginica_u_x2,Iris_virginica_u_x3,Iris_virginica_u_x4,\ Iris_virginica_variance_x1,Iris_virginica_variance_x2,Iris_virginica_variance_x3,\ Iris_virginica_variance_x4=calculate_junzhi_and_fangcha(Iris_virginica_data) '''开始测试''' test_data=read_test_data(testname) #print ('test_data',test_data) #估计类先验概率 p1=len(Iris_setosa_data)/(len(Iris_versicolor_data)+len(Iris_virginica_data)+len(Iris_setosa_data)) p2=len(Iris_versicolor_data)/(len(Iris_versicolor_data)+len(Iris_virginica_data)+len(Iris_setosa_data)) p3=len(Iris_virginica_data)/(len(Iris_versicolor_data)+len(Iris_virginica_data)+len(Iris_setosa_data)) for x in test_data: #在Iris_setosa种类上的各个特征属性的条件概率 P_x1_Iris_setosa,P_x2_Iris_setosa,P_x3_Iris_setosa,P_x4_Iris_setosa=calculate_P_xi_c(Iris_setosa_u_x1,Iris_setosa_u_x2,Iris_setosa_u_x3,Iris_setosa_u_x4,\ Iris_setosa_variance_x1,Iris_setosa_variance_x2,Iris_setosa_variance_x3,Iris_setosa_variance_x4,x) #print(P_x1_Iris_setosa,P_x2_Iris_setosa,P_x3_Iris_setosa,P_x4_Iris_setosa) #在Iris_versicolor种类上的各个特征属性的条件概率 P_x1_Iris_versicolor,P_x2_Iris_versicolor,P_x3_Iris_versicolor,P_x4_Iris_versicolor=calculate_P_xi_c(Iris_versicolor_u_x1,Iris_versicolor_u_x2,Iris_versicolor_u_x3,Iris_versicolor_u_x4,\ Iris_versicolor_variance_x1,Iris_versicolor_variance_x2,Iris_versicolor_variance_x3,Iris_versicolor_variance_x4,x) #print(P_x1_Iris_versicolor,P_x2_Iris_versicolor,P_x3_Iris_versicolor) #在Iris_virginica种类上的各个特征属性的条件概率 P_x1_Iris_virginica,P_x2_Iris_virginica,P_x3_Iris_virginica,P_x4_Iris_virginica=calculate_P_xi_c(Iris_virginica_u_x1,Iris_virginica_u_x2,Iris_virginica_u_x3,Iris_virginica_u_x4,\ Iris_virginica_variance_x1,Iris_virginica_variance_x2,Iris_virginica_variance_x3,Iris_virginica_variance_x4,x) #print(P_x1_Iris_virginica,P_x2_Iris_virginica,P_x3_Iris_virginica,P_x4_Iris_virginica) #计算各个种类上的后验概率 P_Iris_setosa=p1*P_x1_Iris_setosa*P_x2_Iris_setosa*P_x3_Iris_setosa*P_x4_Iris_setosa #print( P_Iris_setosa) P_Iris_versicolor=p2*P_x1_Iris_versicolor*P_x2_Iris_versicolor*P_x3_Iris_versicolor*P_x4_Iris_versicolor #print( P_Iris_versicolor) P_Iris_virginica=p3*P_x1_Iris_virginica*P_x2_Iris_virginica*P_x3_Iris_virginica*P_x4_Iris_virginica #print( P_Iris_virginica) if P_Iris_setosa>P_Iris_versicolor and P_Iris_setosa>P_Iris_virginica: print(x[0],x[1],x[2],x[3],":这行数据属于Iris_setosa类") if P_Iris_versicolor>P_Iris_setosa and P_Iris_versicolor>P_Iris_virginica: print(x[0],x[1],x[2],x[3],":这行数据属于Iris_versicolor类") if P_Iris_virginica>P_Iris_setosa and P_Iris_virginica>P_Iris_versicolor: print(x[0],x[1],x[2],x[3],":这行数据属于Iris_virginica类")具体细节及数据集见我的github:
https://github.com/Erikfather/bayes-python 