机器学习-线性判别分析python实现

西瓜书p60

import numpy as np 

dataset = [[0.697,0.460,1],[0.774,0.376,1],[0.634,0.264,1],[0.608,0.318,1],[0.556,0.215,1],[0.403,0.211,1],[0.481,0.149,1],[0.437,0.211,1],
        [0.666,0.091,0],[0.243,0.267,0],[0.245,0.057,0],[0.343,0.099,0],[0.639,0.161,0],[0.657,0.198,0],[0.360,0.370,0],[0.593,0.042,0]]

def lda(dataset):
	data = np.array(dataset)[:,0:2]
	label = np.array(dataset)[:,-1]
	x0 = data[label==0]
	x1 = data[label==1] 

	m,n = np.shape(data)
	u0 = x0.mean(0).reshape(1,n)   #reshape将一维arr转为二维arr
	u1 = x1.mean(0).reshape(1,n)

	sw = (x1-u1).T.dot(x1-u1)+(x0-u0).T.dot(x0-u0)
	w = (u0-u1).dot(np.linalg.inv(sw))
	#w = np.mat(u0-u1)*np.mat(sw).I
	return w
	
w = lda(dataset)
print(w)

拓展1
拓展2