机器学习笔记：python中使用numpy实现KNN区分电影类型DEMO（爱情片和动作片分类）

偏执的太偏执、 2023-10-17 21:44 14阅读 0赞

import numpy as np  
import operator  
  
  
\# K近邻算法分类爱情片和动作片  
\# 通过打斗镜头次数和接吻镜头次数区分  
  
  
\# 模拟数据 打斗次数 接吻次数  
def init\_data():  
data\_X = np.array(\[\[1,100\],\[2,96\],\[3,93\],\[6,90\],\[80,3\],\[93,2\],\[86,5\]\])  
data\_Y = \['爱情片','爱情片','爱情片','爱情片','动作片','动作片','动作片'\]  
return data\_X,data\_Y  
  
  
\# data: 测试数据  
\# testData: 数据集合  
\# output:测试数据输出  
\# k:取最接近数据的前几个  
def kNN(data, testData, output, k):  
\# 获取测试数据数量  
dataInputRow = testData.shape\[0\]  
\# np.tile 数组沿各个方向复制  
\# 这里将输入数据copy和测试数据数量保持一致，用来计算和测试数据的欧式距离  
reduceData = np.tile(data, (dataInputRow,1)) - testData  
squareData = reduceData \*\* 2  
squareDataSum = squareData.sum(axis = 1)  
distance = squareDataSum \*\* .5  
sortDistance = distance.argsort()  
dataCount = \{\}  
\# 统计排名靠前k数据的爱情片和动作片次数，取次数最高的做为输出  
for i in range(k):  
output\_ = output\[sortDistance\[i\]\]  
dataCount\[output\_\] = dataCount.get(output\_,0) + 1  
sortDataCount = sorted(dataCount.items(), key = operator.itemgetter(1), reverse = True)  
return sortDataCount\[0\]\[0\]  
  
  
if \_\_name\_\_ == '\_\_main\_\_':  
data\_X,data\_Y = init\_data()

print(kNN(\[3,93\], data\_X, data\_Y, 3))

![70][]

[70]: https://img-blog.csdn.net/20180608183435333?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTA0Nzk5ODk=/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70