<PY>kNN

k近邻算法

trainningData=[(2,3,'r'),(3,4,'r'),(3,4.3,'r'),(1,2,'r'),(3,1,'r'),(3,2.1,'r'),(2,2,'r'),(1,3,'r'),(1,1.5,'r'),(4,2,'r')
,(7,3,'g'),(8,2,'g'),(9,2.5,'g'),(9,3,'g'),(8.2,1,'g'),(7.1,3.1,'g'),(6,6,'g'),(7,4,'g'),(8,2.3,'g'),(9,5.2,'g'),(7.5,2.3,'g')]

# import matplotlib.pyplot as plt
# plt.figure()
# [plt.scatter([td[0]],[td[1]],color=td[2]) for td in trainningData]
# plt.show()

distance=lambda x1,x2,p=2:sum([abs(x1[i]-x2[i])**p for i in range(len(x1))])**(1.0/p) if len(x1)==len(x2) else "length not match"

# def distanceDict(testD,traiD):
# distDict={}
# for i in range(len(traiD)):
# distDict[i]=distance(testD,traiD[i][:-1])

kClasses=lambda testD,traiD,k:[traiD[distDict[0]][-1] for distDict in sorted({i:distance(testD,traiD[i][:-1]) for i in range(len(traiD))}.items(),key=lambda x:x[1],reverse=False)[:k]] # distinct,non-repeatable is index
def findClass(testD,traiD,k):
cs=kClasses(testD,traiD,k)
ss=list(set(cs))
counts=[cs.count(c) for c in ss]
return ss[counts.index(max(counts))]

print(findClass((2,1.3),trainningData,k=5))