KNN算法实现 K近邻分类

K近邻分类算法
下面先介绍算法的原理,其实很简单就是欧式距离＋概率
首先介绍下欧式距离
设向量x=[x1,x2,x3…xn]
则x的L2范数 |L2|=$\sqrt{x{x}^T}$
令K=20 这里的K为超参数 K的设置至关重要 它决定了模型的分类准确率
下面以鸢尾花数据集作为示例

from sklearn.metrics import recall_score,precision_score,f1_score,confusion_matrix
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from collections import Counter
import seaborn as sns
import numpy as np


class KNN():
    def __init__(self,k):
        self.k=k
    def distance(self,X,x):
        return np.linalg.norm(X-x,axis=1)

    def fit(self,X,Y):
        self.X=np.array(X) if not isinstance(X,np.ndarray) else X
        self.Y=np.array(Y) if not isinstance(Y,np.ndarray) else Y
        self.labels=set(Y)

    def pred(self,X,isWeighted=False):
        res=[]
        for x in X:
            dis=self.distance(self.X,x)
            vote=np.argsort(dis)[:self.k]#得到k个最近数据下标
            #加权投票
            if isWeighted:
                temp=1/(dis[vote]+1e3)#距离倒数
                weights=temp/temp.sum()#权重归一化
                ct={}
                for (lab,score) in zip(self.Y[vote],dis[vote]*weights):
                    if lab in ct:
                        ct[lab]+=score
                    else:
                        ct[lab]=score
            #等权投票
            else:
                ct=Counter(self.Y[vote])
            res.append(max(ct,key=ct.get))
        return res

    def getEvaluation(self,ytrue,ypred):
        print("精确率",precision_score(ytrue,ypred,average="micro"))
        print("回归率",recall_score(ytrue,ypred,average="micro"))
        print("f1-score",f1_score(ytrue,ypred,average="micro"))
        confusmatrix=confusion_matrix(ytrue,ypred)
        sns.heatmap(confusmatrix,annot=True,cmap="Greens")


iris=load_iris()
X,Y=iris.data,iris.target
Xtrain,Xtest,Ytrain,Ytest=train_test_split(X,Y,test_size=0.3,random_state=42)
knn=KNN(10)
knn.fit(Xtrain,Ytrain)
mypred=knn.pred(Xtest,True)
knn.getEvaluation(Ytest,mypred)