机器学习小记——KNN（K近邻）^_^（一）

1 #获取数据
2 X=np.loadtxt("/Users/galan/py/ML-D/iris.data.txt",delimiter=",",dtype=float,usecols=(0,1,2,3))
3 y=np.loadtxt("/Users/galan/py/ML-D/iris.data.txt",delimiter=",",dtype=str,usecols=(4,))
4 #创建训练数据和测试数据
5 X_train,X_test,y_train,y_test=train_test_split(X,y,train_size=.7)

def eculidean(p,q):
    sumSq=0.0
    #讲差值德平方累加起来
    for i in range(len(p)):
        sumSq+=sum(p[i]-q[i])**2
    #求平方根
    return (sumSq**0.5)

def pearson(x,y):
    n=len(x)
    vals=range(n)
    #简单求和
    sumx=sum([float(x[i]) for i in vals])
    sumy=sum([float(y[i]) for i in vals])
    #求平方和
    sumxSq=sum([x[i]**2.0 for i in vals])
    sumySq=sum([y[i]**2.0 for i in vals])
    #求乘积之和
    pSum=sum([x[i]*y[i] for i in vals])
    #计算皮尔逊评价值
    num=pSum-(sumx*sumy/n)
    den=((sumxSq-pow(sumx,2)/n)*(sumySq-pow(sumy,2)/n))**.5
    if den==0:return 1
    
    r=num/den
    return r

#vect1,vect2位两个一维向量如(1,1)
def getCost(vect1,vect2):
    sum_x=0.0
    sum_y=0.0
    sum_xy=0.0
    for a,b in zip(vect1,vect2):
        sum_xy+=a*b
        sum_x+=a**2
        sum_y+=b**2
        if sum_x==0.0 or sum_y==0.0:
            return None
        else:
            return sum_xy/((sum_x*sum_y)**0.5)

#K值
k=5
#计算所有的欧氏距离组合成字典
Dists={}
for i in range(len(X_train)):
    Dists[eculidean(X_test[0],X_train[i])]=y_train[i]
#排序字典
sortedDist=sorted(Dists.iteritems(),reverse=True,key=lambda x:x[0])[:k]
classCount={}
#寻找最多的类别标签
for i in sortedDist:
    if i[1] in classCount:
        classCount[i[1]]+=1
    else:
        classCount[i[1]]=1
print classCount

#coding:utf-8
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

#获取数据
X=np.loadtxt("./ML-D/iris.data.txt",delimiter=",",dtype=float,usecols=(0,1,2,3))
y=np.loadtxt("./ML-D/iris.data.txt",delimiter=",",dtype=str,usecols=(4,))
#创建训练数据和测试数据
X_train,X_test,y_train,y_test=train_test_split(X,y,train_size=.7)

def eculidean(p,q):
    sumSq=0.0
    #讲差值德平方累加起来
    for i in range(len(p)):
        sumSq+=sum(p-q[i])**2
    #求平方根
    return (sumSq**0.5)

def classify(X_train,X_test,k):
    #计算所有的欧氏距离
    Dists={}
    for i in range(len(X_train)):
        Dists[eculidean(X_test,X_train[i])]=y_train[i]
    #排序字典
    sortedDist=sorted(Dists.iteritems(),reverse=True,key=lambda x:x[0])[:k]
    classCount={}
    #寻找最多的类别标签
    for i in sortedDist:
        if i[1] in classCount:
            classCount[i[1]]+=1
        else:
            classCount[i[1]]=1
    return sorted(classCount.iteritems(),key=lambda x:x[1],reverse=True)

if __name__ == '__main__':
    print "%s的类别为%s"%(X_test[15],classify(X_train,X_test[0],5)[0][0])