手把手教你使用KNN算法（Python实现）

上节我们简单进行了KNN算法的说明，想想假期结束再回味一下！

Knn算法基本原理：

假设我有如下两个数据集：

dataset = {'black':[ [1,2], [2,3], [3,1] ], 'red':[ [6,5], [7,7], [8,6] ] }

另外有一点绿颜色标记（3.5,5.3）， KNN的任务就是判断这个点(下图中的绿点)该划分到哪个组。

KNN分类算法超级简单：只需使用初中所学的两点距离公式（欧拉距离公式），计算绿点到各组的距离，看绿点和哪组更接近。K代表取离绿点最近的k个点，这k个点如果其中属于红点个数占多数，我们就认为绿点应该划分为红组，反之，则划分为黑组。如果有两组数据(如上图)，k值最小应为3（X轴坐标3.5）。

除了K-Nearest Neighbor之外还有其它分组的方法，如Radius-Based Neighbor。此方法后面在做介绍。

实现代码如下：

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

import math import numpy as np from matplotlib import pyplot from collections import Counter import warnings   # k-Nearest Neighbor算法 def k_nearest_neighbors ( data , predict , k = 5 ) :        if len ( data ) >= k :          warnings . warn ( "k is too small" )        # 计算predict点到各点的距离      distances = [ ]      for group in data :          for features in data [ group ] :              #euclidean_distance = np.sqrt(np.sum((np.array(features)-np.array(predict))**2))   # 计算欧拉距离，这个方法没有下面一行代码快              euclidean_distance = np . linalg . norm ( np . array ( features ) - np . array ( predict ) )              distances . append ( [ euclidean_distance , group ] )        sorted_distances = [ i [ 1 ]    for i in sorted ( distances ) ]      top_nearest = sorted_distances [ : k ]        #print(top_nearest)  ['red','black','red']      group_res = Counter ( top_nearest ) . most_common ( 1 ) [ 0 ] [ 0 ]      confidence = Counter ( top_nearest ) . most_common ( 1 ) [ 0 ] [ 1 ] * 1.0 / k      # confidences是对本次分类的确定程度，例如(red,red,red)，(red,red,black)都分为red组，但是前者显的更自信      return group_res , confidence   if __name__ == '__main__' :        dataset = { 'black' : [ [ 1 , 2 ] , [ 2 , 3 ] , [ 3 , 1 ] ] , 'red' : [ [ 6 , 5 ] , [ 7 , 7 ] , [ 8 , 6 ] ] }      new_features = [ 3.5 , 5.2 ]    # 判断这个样本属于哪个组        for i in dataset :          for ii in dataset [ i ] :              pyplot . scatter ( ii [ 0 ] , ii [ 1 ] , s = 50 , color = i )        which_group , confidence = k_nearest_neighbors ( dataset , new_features , k = 3 )      print ( which_group , confidence )        pyplot . scatter ( new_features [ 0 ] , new_features [ 1 ] , s = 100 , color = which_group )        pyplot . show ( )

结果如下所示：

归为红色一类的概率为：0.66666666

我们使用实际数据进行应用

数据集（Breast Cancer）：https://archive.ics.uci.edu/ml/datasets/Breast+Cancer+Wisconsin+%28Original%29

点击download： Data Folder/breast-cancer-wisconsin.data(复制粘贴到txt文件再重命名)

代码如下：（if __name__=='__main__':前面代码一样）

import math

import numpy as np

from collections import Counter

import warnings

import pandas as pd

import random

 

# k-Nearest Neighbor算法

def k_nearest_neighbors ( data , predict , k = 5 ) :

 

     if len ( data ) >= k :

         warnings . warn ( "k is too small" )

 

     # 计算predict点到各点的距离

     distances = [ ]

     for group in data :

         for features in data [ group ] :

             euclidean_distance = np . linalg . norm ( np . array ( features ) - np . array ( predict ) )

             distances . append ( [ euclidean_distance , group ] )

 

     sorted_distances = [ i [ 1 ]    for i in sorted ( distances ) ]

     top_nearest = sorted_distances [ : k ]

 

     group_res = Counter ( top_nearest ) . most_common ( 1 ) [ 0 ] [ 0 ]

     confidence = Counter ( top_nearest ) . most_common ( 1 ) [ 0 ] [ 1 ] * 1.0 / k

 

     return group_res , confidence

if __name__=='__main__':

    df=pd.read_csv('iris.csv')#加载数据

    #print （df.head（））

    #print(df.shape)

     df . replace ( '?' , np . nan , inplace = True )    # -99999

     df . dropna ( inplace = True )    # 去掉无效数据

     #print(df.shape)

     df . drop ( [ 'id' ] , 1 , inplace = True )#去掉id 这一列（第一列名字为id）

 

     # 把数据分成两部分，训练数据和测试数据

     full_data = df . astype ( float ) . values . tolist ( )

 

     random . shuffle ( full_data )

 

     test_size = 0.2    # 测试数据占20%

     train_data = full_data [ : - int ( test_size * len ( full_data ) ) ]

     test_data = full_data [ - int ( test_size * len ( full_data ) ) : ]

 

     train_set = { 2 : [ ] , 4 : [ ] }

     test_set = { 2 : [ ] , 4 : [ ] }

     for i in train_data :

         train_set [ i [ - 1 ] ] . append ( i [ : - 1 ] )

     for i in test_data :

         test_set [ i [ - 1 ] ] . append ( i [ : - 1 ] )

 

     correct = 0

     total = 0

 

     for group in test_set :

         for data in test_set [ group ] :

             res , confidence = k_nearest_neighbors ( train_set , data , k = 5 ) # 你可以调整这个k看看准确率的变化，你也可以使用matplotlib画出k对应的准确率，找到最好的k值

             if group == res :

                 correct += 1

             else :

                 print ( confidence )

             total += 1

 

     print ( correct / total )    # 准确率

 

     print ( k_nearest_neighbors ( train_set , [ 4 , 2 , 1 , 1 , 1 , 2 , 3 , 2 , 1 ] , k = 5 ) ) # 预测一条记录

结果如下所示：

使用scikit-learn 中K临近算法

代码如下：

import numpy as np

from sklearn import preprocessing , cross_validation , neighbors    # cross_validation已deprecated，使用model_selection替代

import pandas as pd

 

df=pd.read_csv('iris.csv')#加载exel数据

#print(df.head())

#print(df.shape)

df . replace ( '?' , np . nan , inplace = True )    # -99999

df . dropna ( inplace = True )

#print(df.shape)

df . drop ( [ 'id' ] , 1 , inplace = True )

 

X = np . array ( df . drop ( [ 'class' ] , 1 ) )

Y = np . array ( df [ 'class' ] )

 

X_trian , X_test , Y_train , Y_test = cross_validation . train_test_split ( X , Y , test_size = 0.2 )

 

clf = neighbors . KNeighborsClassifier ( )

clf . fit ( X_trian , Y_train )

 

accuracy = clf . score ( X_test , Y_test )

print ( accuracy )

 

sample = np . array ( [ 4 , 2 , 1 , 1 , 1 , 2 , 3 , 2 , 1 ] )

print ( sample . reshape ( 1 , - 1 ) )

print ( clf . predict ( sample . reshape ( 1 , - 1 ) ) )

结果如下：（里面有个警告但不妨碍结果）

scikit-learn中的算法和我们上面实现的算法原理完全一样，只是它的效率更高，支持的参数更全。 (以上内容学习于大熊猫)

breast-cancer-wisconsin.data