4.2 最邻近规则分类(K-Nearest Neighbor)KNN算法应用

Posted 2020-08-22 Michael2397

4.2 最邻近规则分类(K-Nearest Neighbor)KNN算法应用

1 数据集介绍：

 

虹膜

 

 

 

150个实例

 

萼片长度，萼片宽度，花瓣长度，花瓣宽度

(sepal length, sepal width, petal length and petal width）

 

类别：

Iris setosa, Iris versicolor, Iris virginica.

 

 

 

 

 

2. 利用Python的机器学习库sklearn: SkLearnExample.py

 

from sklearn import neighbors

from sklearn import datasets

 

knn = neighbors.KNeighborsClassifier()

 

 

iris = datasets.load_iris()

 

 

print iris

 

knn.fit(iris.data, iris.target)

 

predictedLabel = knn.predict([[0.1, 0.2, 0.3, 0.4]])

 

print predictedLabel

 

 

 

 

3. KNN 实现Implementation:

 

 

# Example of kNN implemented from Scratch in Python

 

import csv

import random

import math

import operator

 

def loadDataset(filename, split, trainingSet=[] , testSet=[]):

    with open(filename, \'rb\') as csvfile:

        lines = csv.reader(csvfile)

        dataset = list(lines)

        for x in range(len(dataset)-1):

            for y in range(4):

                dataset[x][y] = float(dataset[x][y])

            if random.random() < split:

                trainingSet.append(dataset[x])

            else:

                testSet.append(dataset[x])

 

 

def euclideanDistance(instance1, instance2, length):

    distance = 0

    for x in range(length):

        distance += pow((instance1[x] - instance2[x]), 2)

    return math.sqrt(distance)

 

def getNeighbors(trainingSet, testInstance, k):

    distances = []

    length = len(testInstance)-1

    for x in range(len(trainingSet)):

        dist = euclideanDistance(testInstance, trainingSet[x], length)

        distances.append((trainingSet[x], dist))

    distances.sort(key=operator.itemgetter(1))

    neighbors = []

    for x in range(k):

        neighbors.append(distances[x][0])

    return neighbors

 

def getResponse(neighbors):

    classVotes = {}

    for x in range(len(neighbors)):

        response = neighbors[x][-1]

        if response in classVotes:

            classVotes[response] += 1

        else:

            classVotes[response] = 1

    sortedVotes = sorted(classVotes.iteritems(), key=operator.itemgetter(1), reverse=True)

    return sortedVotes[0][0]

 

def getAccuracy(testSet, predictions):

    correct = 0

    for x in range(len(testSet)):

        if testSet[x][-1] == predictions[x]:

            correct += 1

    return (correct/float(len(testSet))) * 100.0

    

def main():

    # prepare data

    trainingSet=[]

    testSet=[]

    split = 0.67

    loadDataset(r\'D:\\MaiziEdu\\DeepLearningBasics_MachineLearning\\Datasets\\iris.data.txt\', split, trainingSet, testSet)

    print \'Train set: \' + repr(len(trainingSet))

    print \'Test set: \' + repr(len(testSet))

    # generate predictions

    predictions=[]

    k = 3

    for x in range(len(testSet)):

        neighbors = getNeighbors(trainingSet, testSet[x], k)

        result = getResponse(neighbors)

        predictions.append(result)

        print(\'> predicted=\' + repr(result) + \', actual=\' + repr(testSet[x][-1]))

    accuracy = getAccuracy(testSet, predictions)

    print(\'Accuracy: \' + repr(accuracy) + \'%\')

    

main()