阿里云可以做网站,asp网站建设项目实训,渠道网创业加盟博览会,如何制作自己的网站二维码本文使用的测试问题是“皮马印第安人糖尿病问题”这个问题包括768个对于皮马印第安患者的医疗观测细节#xff0c;记录所描述的瞬时测量取自患者的年龄#xff0c;怀孕和血液检查的次数。所有患者都是21岁以上的女性#xff0c;所有属性都是数值型#xff0c;而且属性的单位…本文使用的测试问题是“皮马印第安人糖尿病问题”这个问题包括768个对于皮马印第安患者的医疗观测细节记录所描述的瞬时测量取自患者的年龄怀孕和血液检查的次数。所有患者都是21岁以上的女性所有属性都是数值型而且属性的单位各不相同每一个记录归属一个类最后一列为类标签表示患者是否在5年之内感染的糖尿病。如果是则为1否则为0下面是文件的地址以.csv扩展名保存第一次写如有不足请多多包涵# -*- coding: utf-8 -*-Created on Wed Mar 29 10:49:47 2017author: Administratorimport csvimport numpy as npimport math#处理数据data_namedata.csvnp.random.seed(2017)#训练集80% 测试集20%def splitData(filename):testSets[]trainSets[]lineslist(csv.reader(open(filename,r)))for line in lines:pnp.random.randint(100)if p20:testSets.append(line)else:trainSets.append(line)return trainSets,testSetstrainSets,testSetssplitData(data_name)#分为特征和类别def fea_and_class(trs,tes):tr_feas0[]tr_feas1[]tr_dict{}te_feas0[]te_feas1[]te_dict{}for tr in trs:tr_fea[float(x) for x in tr[:8]]tr_clint(tr[-1])if tr_cl0:#tr_dict[tr_cl]tr_featr_feas0.append(tr_fea)if tr_cl1:tr_feas1.append(tr_fea)tr_dict[0]tr_feas0tr_dict[1]tr_feas1for te in tes:te_fea[float(x) for x in te[:8]] #前八个为特征最后一个为类别标签te_clint(te[-1])#te_dict[te_cl]te_feaif te_cl0:te_feas0.append(te_fea)if te_cl1:te_feas1.append(te_fea)te_dict[0]te_feas0te_dict[1]te_feas1return tr_dict,te_dict#return tr_dict,te_dicttr_dict,te_dictfea_and_class(trainSets,testSets)#提取训练集的属性特征trDict{}tr[]te[]for i in range(len(tr_class)):if tr_class[i][0]0:tr.append(trSets[i])if tr_class[i][0]1:te.append(trSets[i])trDict[0]trtrDict[1]te#训练集的同一类的均值方差tr_mean0np.mean(tr_dict[0],axis0)tr_var0np.var(tr_dict[0],axis0)tr_mean1np.mean(tr_dict[1],axis0)tr_var1np.var(tr_dict[1],axis0)#假设数据服从高斯分布def gaussian(x,mu,sigma):val1/math.sqrt(2*math.pi*sigma)return val*(math.exp((-(x-mu)**2)/(2*sigma)))#valuesgaussian(10.0,tr_mean0[0],tr_var0[0])pre_cla0[]pre_cla1[]#对于测试集判断其属于哪些类并计算准确率假设特征独立同分布for te0 in te_dict[0]:val01val11for i in range(len(te0)):val0gaussian(te0[i],tr_mean0[i],tr_var0[i])val0*val0val1gaussian(te0[i],tr_mean1[i],tr_var1[i])val1*val1if val0val1:classes0else:classes1pre_cla0.append(classes)for te1 in te_dict[1]:val0_1val1_1for i in range(len(te1)):val0_gaussian(te1[i],tr_mean0[i],tr_var0[i])val0_*val0_val1_gaussian(te1[i],tr_mean1[i],tr_var1[i])val1_*val1_if val0_val1_:classes0else:classes1pre_cla1.append(classes)count0for pre_cla0_each in pre_cla0:if pre_cla0_each0:count1for pre_cla1_each in pre_cla1:if pre_cla1_each1:count1acccount/(len(pre_cla0)len(pre_cla1))
