数据挖掘期末论文要领.pdf

KNN分类算法的研究 KNN KNN分类算法的研究 KKNNNN 吴雪莲 1.浙江大学宁波理工学院 宁波, 315000 E-mail:AndreMouche@126.com 摘 要:KNN算法是应用最广泛的分类技术之一。本文简要介绍了KNN算法的基本原理，重点介 绍了KNN算法中K值，投票函数对分类的影响，并研究各个属性对分类的贡献。文章最后提出一 种改进的KNN算法，该算法综合考虑KNN算法中各个参数对分类的影响，建立的分类模型相比 于原有的算法对噪声的敏感度降低，准确率也相对较高。此外，自编程序实现了以上分析实验中 用到的算法，并给出了相关实验结果。 关键词:KNN,K,投票函数，改进，分类。 ResearchonK-nearestNeighborAlgorithminClassification ResearchonK-nearestNeighborAlgorithminClassification RReesseeaarrcchhoonnKK--nneeaarreessttNNeeiigghhbboorrAAllggoorriitthhmmiinnCCllaassssiiffiiccaattiioonn WuXuelian NingboInstituteOfTechnology,ZhejiangUniversity,ningbo 315000 E-mail:AndreMouche@126.com Abstract: Abstract: AAbbssttrraacctt:: K-Nearest Neighbor Algorithm is one of the most widely used classification technologies. This article introduced the fundamental theory of KNN algorithm briefly, and focused on the introduction of the influence which may caused bythe parameter K,vote approaches ,also,weresearchthe different attributions contribution to the result of classification . At last we proposed one kind of algorithm based on improved KNN algorithm,which has considered all those parameters which would influence the result of the classification.TheimprovedKNN has highaccuracywithlowsensitivity tonoise.We provide allthe programs whichused inthe experiments andtheresultof alltheexperiments. ， Keywords: Keywords: KKeeyywwoorrddss::KNN,K,vote approach,improved,classification