KNN算法和實現

 

 KNN要用到歐氏距離

 

 KNN下面的缺點很容易使分類出錯（比如下面黑色的點）

 

 下面是KNN算法的三個例子demo，

第一個例子是根據算法原理實現

 

import matplotlib.pyplot as plt
import numpy as np
import operator
# 已知分類的數據
x1 = np.array([3,2,1])
y1 = np.array([104,100,81])
x2 = np.array([101,99,98])
y2 = np.array([10,5,2])
scatter1 = plt.scatter(x1,y1,c='r')
scatter2 = plt.scatter(x2,y2,c='b')
# 未知數據
x = np.array([18])
y = np.array([90])
scatter3 = plt.scatter(x,y,c='k')
#畫圖例
plt.legend(handles=[scatter1,scatter2,scatter3],labels=['labelA','labelB','X'],loc='best')
plt.show()
# 已知分類的數據
x_data = np.array([[3,104],
                   [2,100],
                   [1,81],
                   [101,10],
                   [99,5],
                   [81,2]])
y_data = np.array(['A','A','A','B','B','B'])
x_test = np.array([18,90])
# 計算樣本數量
x_data_size = x_data.shape[0]
print(x_data_size)
# 復制x_test
print(np.tile(x_test, (x_data_size,1)))
# 計算x_test與每一個樣本的差值
diffMat = np.tile(x_test, (x_data_size,1)) - x_data
diffMat
# 計算差值的平方
sqDiffMat = diffMat**2
sqDiffMat
# 求和
sqDistances = sqDiffMat.sum(axis=1)
sqDistances
# 開方
distances = sqDistances**0.5
print(distances)
# 從小到大排序
sortedDistances = distances.argsort()#返回distances里的數據從小到大的下標數組
print(sortedDistances)
classCount = {}
# 設置k
k = 5
for i in range(k):
    # 獲取標簽
    votelabel = y_data[sortedDistances[i]]
    # 統計標簽數量
    classCount[votelabel] = classCount.get(votelabel,0) + 1#)0表示沒有該字典里沒有該值時默認為0
classCount
# 根據operator.itemgetter(1)-第1個值對classCount排序，然后再取倒序
sortedClassCount = sorted(classCount.items(),key=operator.itemgetter(1), reverse=True)
print(sortedClassCount)
# 獲取數量最多的標簽
knnclass = sortedClassCount[0][0]#第一個0表示取第一個鍵值對('A', 3)，第二個0表示取('A', 3)的‘A’
print(knnclass)

 1 import numpy as np#對iris數據集進行訓練分類
from sklearn import  datasets
from sklearn.model_selection import  train_test_split
from sklearn.metrics import  classification_report,confusion_matrix#對模型分類結果進行評估的兩個模型
import operator#https://blog.csdn.net/u010339879/article/details/98304292，關於operator的使用
import  random
def knn(x_test, x_data, y_data, k):
    x_data_size = x_data.shape[0]  # 計算樣本數量
    diffMat = np.tile(x_test,(x_data_size,1)) - x_data# 復制x_test，計算x_test與每一個樣本的差值
    sqDiffMat = diffMat**2# # 計算差值的平方
    sqDistance = sqDiffMat.sum(axis= 1) # 求和
    distances = sqDistance**0.5 # 開方
    sortedDistance = distances.argsort()# 從小到大排序
    classCount = {}
    for i in range(k):
        vlabel = y_data[sortedDistance[i]] # 獲取標簽
        classCount[vlabel] = classCount.get(vlabel,0)+1# 統計標簽數量
    sortedClassCount = sorted(classCount.items(),key = operator.itemgetter(1), reverse = True) # 根據operator.itemgetter(1)-第1個值對classCount排序，然后再取倒序
    return  sortedClassCount[0][0]
iris = datasets.load_iris()# 載入數據
x_train,x_test,y_train,y_test = train_test_split(iris.data, iris.target, test_size=0.3)
#打亂數據
# data_size = iris.data.shape[0]
# index = [i for i in range(data_size)]
# random.shuffle(index)
# iris.data = iris.data[index]
# iris.target = iris.target[index]
# test_size = 40#切分數據集
# x_train = iris.data[test_size:]
# x_test = iris.data[:test_size]
# y_train = iris.target[test_size:]
# y_test = iris.target[:test_size]
prodictions = []
for i in range(x_test.shape[0]):
    prodictions.append(knn(x_test[i],x_train,y_train,5))
print(prodictions)
print(classification_report(y_test, prodictions))
print(confusion_matrix(y_test,prodictions))
#關於混淆矩陣可以看這篇博客，#https://www.cnblogs.com/missidiot/p/9450662.html

 

# 導入算法包以及數據集
from sklearn import neighbors
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
import random
# 載入數據
iris = datasets.load_iris()
#print(iris)
# 打亂數據切分數據集
# x_train,x_test,y_train,y_test = train_test_split(iris.data, iris.target, test_size=0.2) #分割數據0.2為測試數據，0.8為訓練數據

#打亂數據
data_size = iris.data.shape[0]
index = [i for i in range(data_size)]
random.shuffle(index)
iris.data = iris.data[index]
iris.target = iris.target[index]

#切分數據集
test_size = 40
x_train = iris.data[test_size:]
x_test =  iris.data[:test_size]
y_train = iris.target[test_size:]
y_test = iris.target[:test_size]

# 構建模型
model = neighbors.KNeighborsClassifier(n_neighbors=3)
model.fit(x_train, y_train)
prediction = model.predict(x_test)
print(prediction)
print(classification_report(y_test, prediction))

這三個代碼第一個，第二個是根據底層原理實現knn算法，第三個則是調用庫函數處理數據。

 下面一個代碼是利用第三個代碼中用到的庫實現第一個代碼功能，可以發現使用系統提供的庫，簡單許多

from sklearn import  neighbors
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
import numpy as np
x_data = np.array([[3,104],
                   [2,100],
                   [1,81],
                   [101,10],
                   [99,5],
                   [81,2]])
y_data = np.array(['A','A','A','B','B','B'])
x_test1 = np.array([[18,90]])
x_train, x_test, y_train,y_test = train_test_split(x_data, y_data,test_size= 0.3)
model = neighbors.KNeighborsClassifier(n_neighbors=3)
model.fit(x_train, y_train)
print(x_test1)
prediction = model.predict(x_test1)
print(prediction)