利用python实现K近邻算法

总目录：Python数据分析整理

本文数据以及大部分代码来自《机器学习实战》

机器学习实战

---

数据集

https://www.manning.com/downloads/1108
\machinelearninginaction\Ch02\datingTestSet.txt
\machinelearninginaction\Ch02\datingTestSet2.txt

KNN.py

from numpy import *
import operator

import pandas as pd

# k近临算法实现函数
# dataSet=数据集
# labels=分类的值
# k=规定周围点的个数
# 函数返回该点分类结果
def classify0(inX, dataSet, labels, k):
    dataSetSize = dataSet.shape[0]
    diffMat = tile(inX, (dataSetSize,1)) - dataSet
    sqDiffMat = diffMat**2
    sqDistances = sqDiffMat.sum(axis=1)
    distances = sqDistances**0.5
    sortedDistIndicies = distances.argsort()
    classCount={
    
    }
    for i in range(k):
        voteIlabel = labels[sortedDistIndicies[i]]
        classCount[voteIlabel] = classCount.get(voteIlabel,0) + 1
    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)
    # print(sortedClassCount)
    return sortedClassCount[0][0]

# 数组标准化函数
# 输入原始数据
# 输出标准化的数据，并输出原始数据的范围
def autoNorm(dataSet):
    minVals = dataSet.min(0)
    maxVals = dataSet.max(0)
    ranges = maxVals - minVals
    # normDataSet = zeros(shape(dataSet))
    m, n = dataSet.shape
    normDataSet = dataSet - tile(minVals, (m,1))
    normDataSet = normDataSet/tile(ranges, (m,1))   #element wise divide
    b = pd.DataFrame(columns=['min'], data=minVals)
    b['ranges'] = ranges
    b['max'] = maxVals
    return normDataSet, b

# 拆分原始数据，与训练数据的x,y
def split_data(train,test):
    train_x = train[[0,1,2]]
    train_y = train[[3]]
    test_x = test[[0, 1, 2]]
    test_y = test[[3]]
    return train_x,train_y,test_x,test_y

数据分析

from k近邻算法 import KNN
import pandas as pd
import matplotlib.pyplot as plt

plt.rcParams['font.sans-serif'] = ['SimHei']

data_file = pd.read_csv('datingTestSet2.txt', sep='\t', header=None)

# didntLike=不喜欢=1
# smallDoses=魅力一般=2
# largeDoses=极具魅力=3
data1 = data_file[data_file[3]==1]
data2 = data_file[data_file[3]==2]
data3 = data_file[data_file[3]==3]
plt.scatter(data1[0], data1[1], c='r', alpha=0.4, label='不喜欢')
plt.scatter(data2[0], data2[1], c='b', alpha=0.4, label='魅力一般')
plt.scatter(data3[0], data3[1], c='g', alpha=0.4, label='极具魅力')
plt.legend(loc=2)
plt.title('12列分类情况')
plt.savefig('12数据分布情况')
plt.show()

plt.scatter(data1[0], data1[2], c='r', alpha=0.4, label='不喜欢')
plt.scatter(data2[0], data2[2], c='b', alpha=0.4, label='魅力一般')
plt.scatter(data3[0], data3[2], c='g', alpha=0.4, label='极具魅力')
plt.legend(loc=2)
plt.title('13列分类情况')
plt.savefig('13数据分布情况')
plt.show()

plt.scatter(data1[1], data1[2], c='r', alpha=0.4, label='不喜欢')
plt.scatter(data2[1], data2[2], c='b', alpha=0.4, label='魅力一般')
plt.scatter(data3[1], data3[2], c='g', alpha=0.4, label='极具魅力')
plt.legend(loc=2)
plt.title('23列分类情况')
plt.savefig('23数据分布情况')
plt.show()

K近邻分类

from k近邻算法 import KNN
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
from numpy import *

plt.rcParams['font.sans-serif'] = ['SimHei']

train_file = pd.read_csv('datingTestSet2.txt', sep='\t', header=None)
test_file = pd.read_csv('datingTestSet.txt', sep='\t', header=None)

# 缺失值检测
print(train_file[train_file.isnull() == True].count())
print(test_file[test_file.isnull() == True].count())

# 检查数据类型
print(train_file.info())
print(test_file.info())

# 将测试集的文字转换为数字
# didntLike=不喜欢=1
# smallDoses=魅力一般=2
# largeDoses=极具魅力=3
test_file.replace({
    
    'didntLike':1,'smallDoses':2,'largeDoses':3}, inplace=True)

# 数据归一化
new_train_file, train_info = KNN.autoNorm(train_file)
new_test_file, test_info = KNN.autoNorm(test_file)
print(train_info)
print(test_info)

# 具体拆分数据
train_x,train_y,test_x,test_y = KNN.split_data(new_train_file, new_test_file)
print(train_x)
print(train_y)
print(test_x)
print(test_y)

lines, columns = new_test_file.shape

def test_knn(k):
    error = 0
    for i in range(lines):
        pred = KNN.classify0(test_x[:][i:i+1].values[0], train_x.values, train_y.T.values[0],k)
        real = test_y[:][i:i+1].values[0][0]
        if real != pred:
            error = error+1
    return error/lines

for i in range(1,6):
    rate = test_knn(i)
    print('k取值：', i, '错误率：', rate)

输出结果：

k取值： 1 错误率： 0.0
k取值： 2 错误率： 0.0
k取值： 3 错误率： 0.027
k取值： 4 错误率： 0.02
k取值： 5 错误率： 0.036