摘要
为了充分掌握品牌市场发展状况,必须建立一套完整的品牌市场发展状况判别体系。文章采用NB分类算法与随机森林法,分别构建了判别模型进行品牌状态判别,对两个模型性能进行分析对比,发现均有不足。为此,利用差异化权值法建立两种模型的组合判别模型,即CM判别模型。最后将三个模型应用于实际场景,结果表明CM判别模型更加符合实际。
In order to fully understand the market development status of the brand, it's necessary to establish a complete system for judging the development status. In this paper, the NB classification algorithm and the random forest method are used respectively to construct the model to judge the brand status. The performance of the two models is analyzed and compared. To this end, the differentiated weighting method is used to establish a combined discriminant model with the two models, namely the CM discriminant model. Finally, the three models are applied to the actual scene, and the results show that the CM discriminant model is more realistic.
引文
[1]吴诚堃.数据挖掘与分析概念与算法[M].北京:人民邮电出版社,2017,09.
[2]杨忠强.基于属性加权和归约的朴素贝叶斯算法研究[D].广西大学,2013.
[3]白赞.基于属性选择加权的朴素贝叶斯算法的改进与应用[D].西安理工大学,2017.
[4]张永潘.基于大数据平台的决策树分类算法及并行化研究[D].南京邮电大学,2017.
[5]张永,丁超,安海岗,等.数据挖掘在电子商务领域中的应用[M].北京:冶金工业出版社,2015.
[6]马晓东.基于加权决策树的随机森林模型优化[D].华中师范大学,2017.
[7]李洪成,陈道轮,吴立明.数据挖掘与R语言[M].北京:机械工业出版,2013.
[8]童先群.基于属性值信息熵的KNN改进算法[J].计算机工程与应用,2010,46(3):115.