python对西瓜数据集进行线性判别，并画图

import matplotlib.pyplot as plt import numpy as np import xlrd table = xlrd.open_workbook('xian.xlsx').sheets()[0]#读取Excel数据 data = [] for i in range(1,table.nrows):#假设第一行是表头不读入 data.append(table.row_values(i)) class0 = [] class1 = [] #划分正反特征集，编号第一列，类别最后一列，特征在中间 for i in data: if i[-1] == 0: class0.append(i[1:-1]) else: class1.append(i[1:-1]) data = np.array(data) #转为数字矩阵 class0 = np.array(class0) #特征都是行向量，组成矩阵 class1 = np.array(class1) # %% #计算相应类别特征的平均 n0 = len(class0) n1 = len(class1) miu0 = np.dot(np.ones([1,n0]),class0)/n0 miu1 = np.dot(np.ones([1,n1]),class1)/n1 #%% #计算类内散度矩阵 s0 = class0 - miu0 s1 = class1 - miu1 Sw = np.dot(s0.transpose(),s0)+np.dot(s1.transpose(),s1) W = np.dot(np.linalg.pinv(Sw),(miu0-miu1).transpose()) #计算W #输出W、miu0和miu1在映射后的值 miu0_LDA = np.dot(miu0,W) miu1_LDA = np.dot(miu1,W) print("变换向量W：") print(W) print("0类的LDA均值："+str(miu0_LDA[0,0])) print("1类的LDA均值："+str(miu1_LDA[0,0])) #%% #判断类别 c_discrim = np.dot(data[:,1:-1],W) #统计正确率 right = 0 for i in range(len(data)): if np.abs(miu0_LDA[0,0] - c_discrim[i]) < np.abs(miu1_LDA[0,0] - c_discrim[i]): if data[i][-1] == 0: right +=1 else: if data[i][-1] == 1: right +=1 print("正确率："+str(right / len(data))) #%% #画图(仅适用于二维特征) ##################图一 fig = plt.figure() ax = fig.add_subplot(121,projection = '3d') plt.xlabel("Feature 1") plt.ylabel("Feature 2") ax.plot(class0[:,0],class0[:,1],'o',label = 'Class0',color = "red") #0类 ax.plot([miu0[0,0]],[miu0[0,1]],'*',label = 'Class0 average',color = "black",markersize = 10) #0类平均 ax.plot(class1[:,0],class1[:,1],'o',label = 'Class1',color = "blue") #1类 ax.plot([miu1[0,0]],[miu1[0,1]],'*',label = 'Class1 average',color = "green",markersize = 10) #1类平均 #映射平面 t = np.linspace(-5,10,10) X,Y = np.meshgrid(t,t) ax.plot_surface(X,Y,X*W[0]+Y*W[1],alpha = 0.5) ax.legend(loc = 'upper left') ##################图二 ax = fig.add_subplot(122,projection = '3d') plt.xlabel("Feature 1") plt.ylabel("Feature 2") ax.plot(class0[:,0],class0[:,1],np.dot(class0,W)[:,0],'o',label = 'Mapping Class0',color = "red") #0类映射 ax.plot([miu0[0,0]],[miu0[0,1]],np.dot(miu0[0],W),'*',label = 'Mapping class0 average',color = "black",markersize = 10) #0类平均映射 ax.plot(class1[:,0],class1[:,1],np.dot(class1,W)[:,0],'o',label = 'Mapping Class1',color = "blue") #1类映射 ax.plot([miu1[0,0]],[miu1[0,1]],np.dot(miu1[0],W),'*',label = 'Mapping class1 average',color = "green",markersize = 10) #1类平均映射 ax.plot(np.zeros([len(class0)]),np.zeros([len(class0)]),np.dot(class0,W)[:,0],'o',color = "red",alpha = 0.5) #0类映射值 ax.plot(np.zeros([len(class1)]),np.zeros([len(class1)]),np.dot(class1,W)[:,0],'o',color = "blue",alpha = 0.5) #1类映射值 ax.plot([np.zeros([1])],np.zeros([1]),np.dot(miu0[0],W),'*',color = "black",alpha = 0.5,markersize = 10) #0类平均映射值 ax.plot([np.zeros([1])],np.zeros([1]),np.dot(miu1[0],W),'*',color = "green",alpha = 0.5,markersize = 10) #1类平均映射值 #映射平面 X,Y = np.meshgrid(t,t) ax.plot_surface(X,Y,X*W[0]+Y*W[1],alpha = 0.5) ax.legend(loc = 'upper left') plt.show()