基于改进RFM模型的电商用户价值分类的研究

Research on E-commerce User Value Classification Based on Improved RFM Model

互联网行业数据量的不断积累膨胀,使得企业内保存了大量的原始数据和各种业务数据。如何有效地对这些数据加以利用,并构建出精准的用户画像系统是每个电商企业发展战略中的重要一环,能够帮助企业更准确地洞察用户的需求,而RFM模型对用户价值的精确分类又是构建用户画像系统的关键。该文提出了一种基于改进的RFM模型的电商用户价值分类方法。该方法通过增加对用户分析的指标,先利用层次分析法对指标进行两两打分、构成判断矩阵、确定各个指标的权重值,然后以肘部法则和轮廓系数为标准,选择出聚类效果最佳的K值,用K-Means‖聚类算法对用户价值进行分类。最后,通过对某电商用户行为数据集进行实验并与其他RFM模型进行对比实验,结果分析表明,该RFME模型与传统的RFM模型和一些改进的RFM模型相比轮廓系数更高,价值分类体系更加客观,具有更精确、更快速的用户分类效果。

With the continuous accumulation and expansion of data in the Internet industry,a large number of original data and various business data are saved in the enterprise.How to effectively use these data and build an accurate user portrait system is an important part of the development strategy of each e-commerce enterprise,which can help enterprises have a more accurate insight into the needs of users,and the accurate classification of user value by RFM model is the key to building a user portrait system.We propose an e-commerce user value classification method based on the improved RFM model.By adding the indicators for user analysis,the proposed method first scores the indicators in pairs by using the analytic hierarchy process,forms a judgment matrix,determines the weight value of each indicator,and then selects the K value with the best clustering effect by using the elbow rule and contour coefficient as the standard.K-means‖clustering algorithm is used to classify user value.Finally,the experiment of an e-commerce user behavior data set and the comparison experiment with other RFM models show that the RFME model has higher contour coefficient,more objective value classification system and more accurate and faster user classification effect than the traditional RFM model and some improved RFM models.