基于西瓜資料集的KNN演算法實作

因為本人在學習這塊內容之后，發現網路上大部分現有代碼的不簡潔以及運行報錯，再者想要的表達方法的不同，所以自己動手結合網路上已有的代碼改寫了一個，運行正常，
代碼及資料集以上傳到GitHub：https://github.com/zhurui-king/aaa

# -*- coding:utf-8 -*-
# Author: 非魚子焉
# Creation_time: 2020.11.11
# Content: 基于西瓜資料集的KNN演算法實作
# Blog: https://zhu-rui.blog.csdn.net/
# GitHub: https://github.com/zhurui-king/aaa

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# KNN演算法類
class KNN(object):
    def __init__(self, x, y, K):#x:密度；y:含糖率；k:近鄰數
        self.x = x
        self.y = y
        self.K = K
        self.n = len(x)
        
	# 計算距離
    def distance(self, p1, p2):
        return np.linalg.norm(np.array(p1) - np.array(p2))
        
	#演算法實作
    def knn(self, x):
        distance = []
        for i in range(self.n):
            dist = self.distance(x, self.x[i])
            distance.append([self.x[i], self.y[i], dist])
        distance.sort(key=lambda x: x[2])
        neighbors = []
        neighbors_labels = []
        for k in range(self.K):
            neighbors.append(distance[k][0])  # 近鄰具體資料
            neighbors_labels.append(distance[k][1])  # 近鄰標記
        return neighbors, neighbors_labels
        
	#選擇多數投票數
    def vote(self, x):
        neighbors, neighbors_labels = self.knn(x)
        vote = {}  # 投票法
        for label in neighbors_labels:
            vote[label] = vote.get(label, 0) + 1
        sort_vote = sorted(vote.items(), key=lambda x:x[1], reverse=True)
        return sort_vote[0][0]  # 回傳投票數最多的標記
        
	#對應標記
    def fit(self):
        labels = []
        for sample in self.x:
            label = self.vote(sample)
            labels.append(label)
        return labels  # 回傳所有樣本的標記
        
	# 計算正確率
    def accuracy(self):
        predict_labels = self.fit()
        real_labels = self.y
        correct = 0
        for predict, real in zip(predict_labels, real_labels):
            if int(predict) == int(real):
                correct += 1
        return correct / self.n
        
#讀取資料
def getdata(path):
    dataSet = pd.read_csv(path, delimiter=",")
    X = dataSet[['density', 'sugar_rate']].values
    Y = dataSet['label']
    return X,Y
    
# 進行繪圖
def drawpictures(x_positive, y_positive,x_negative, y_negative): 
    plt.scatter(x_positive, y_positive, marker='o', color='red', label='1')
    plt.scatter(x_negative, y_negative, marker='o', color='blue', label='0')
    plt.xlabel('密度')
    plt.ylabel('含糖率')
    plt.rcParams['font.sans-serif'] = ['SimHei']  # 用來正常顯示中文標簽
    plt.legend(loc='upper left')
    plt.show()
    
#訓練資料
def train(X,Y):
    for k in range(1, 9):
        print("*****第%d次*****" %k)
        print('本次knn的k值選取為{}'.format(k))
        knn = KNN(X, Y, k)
        predict = knn.fit()
        print('本次knn的正確率為{}'.format(knn.accuracy()))
        x_positive = []
        y_positive = []
        x_negative = []
        y_negative = []
        for i in range(len(X)):
            if int(predict[i]) == 1:
                x_positive.append(X[i][0])
                y_positive.append(X[i][1])
            else:
                x_negative.append(X[i][0])
                y_negative.append(X[i][1])
        drawpictures(x_positive, y_positive,x_negative, y_negative)

if __name__ == '__main__':
    X,Y = getdata('watermelon3_0a.csv')
    train(X,Y)
    print("************程式運行結束************")

最終結果輸出為所選擇近鄰K的對應的正確率，并且進行plot可視化，其中回圈遍歷每一次的K值（K從1開始到所設定的值-1為止）

參考博文：https://blog.csdn.net/weixin_42152526/article/details/93528560
參考書籍：MACHINE LEARNING 機器學習（周志華）清華大學出版社