基于改进K-近邻规则的数据库营销分析

Analysis of Database Marketing Based on Improved K-nearest Neighbors Rule

文章提出一种基于改进K-近邻规则的数据库营销模型。根据数据样本的近邻信息动态确定其最优的近邻个数K,以避免人为设定K这一重要参数可能造成的影响。根据K个近邻距样本的距离,设定不同的投票权重以预测样本属于某一类别的概率。在实际数据集上的实证结果表明,提出的改进K-近邻规则不仅为K值设定提供了一种有效的方法,还能够提高数据库营销的准确性和结果的可解释性,可以有效应用于实际的数据库营销。

This paper proposes a database marketing model based on improved K-nearest neighbors rule. Optimal number of neighbor K is determined according to the dynamic condition of neighborhood information so as to avoid the possible impact of human intervention on determining the important parameter K. According to the distance between K neighboring samples, the voting weights of different neighbors are set to obtain the probability of predicting a sample belonging to a certain class. The empirical study on a real-world dataset indicates that the proposed approach not only provides an effective way to determine the optimal K value in K-nearest neighbors rule, but also improves the accuracy and interpretability of database marketing, which can be effectively applied to actual database marketing