【无标题】kNN算法学习

import numpy as np
import pandas as pd

class KNN(object):
    def __init__(self,path,testdata,k):
        self.path=path
        self.testData=testdata
        self.k=k

    def readData(self):
        #读取数据
        try:
            self.data = pd.read_csv(self.path,sep=',',encoding='gbk')
        except:
            self.data = pd.read_excel(self.path)
        # return self.data

    def clearData(self):
        #清洗空值的数据
        self.data.dropna(axis=1,how='all',inplace=True)
        self.data.dropna(axis=0,how='all',inplace=True)

    def distance(self):
        # 计算距离
        self.data['distance'] = np.sqrt((self.data['搞笑镜头']-self.testData[0])**2
            +(self.data['拥抱镜头']-self.testData[1])**2+(self.data['打斗镜头']-self.testData[2])**2)

    def KNN_way(self):
        # KNN算法函数
        sort_list = sorted(self.data['distance'])[:self.k]   #将距离一列进行排列存入sort_list
        list=[]
        for i in sort_list:
            list.append(self.data.loc[self.data['distance']==i]['电影类型'].values)
        df = pd.DataFrame({'type':list})

        top_type = df['type'].describe()[2][0]
        print(top_type)

test = KNN(r'C:\Users\Administrator\Desktop\电影分类数据.xlsx',[23, 3, 17],5)
test.readData()
test.clearData()
test.distance()
test.KNN_way()

from sklearn.neighbors import KNeighborsClassifier


knn = neighbors.KNeighborsClassifier()  # 取得knn分类器
knn.fit(data, labels)  # 导入数据进行训练，data为属性，labels为标签
print('预测类型为:', knn.predict([[23，3，17]]))