Research on Customer Value Based on Improved SOM+K-means Algorithm
To improve clustering similarity for multiple feature parameters,an improved clustering algorithm is proposed to address the issues of correlation and unequal distribution of multiple feature parameters,and the RFM customer value model is studied using this algorithm.This improved algorithm constructs a distance function for clustering similarity targets through matrix rotation and compression transformation,as well as covariance matrix processing;Based on this distance function and combining the advantages of SOM algorithm and K-means algorithm,an improved SOM+K-means combination clustering algorithm was designed.Create an RFM customer value model using this algorithm and conduct experimental verification.Through the evaluation of the contour coefficient method,the contour coefficients of the clustering algorithm are improved by approximately 0.129 and 0.126 compared to the original K-means and SOM algorithms,respectively.This clustering algorithm improves the clustering effect of RFM customer value and provides a new clustering method for customer value research.
covariance matrixself organizing neural networkK-meansclustering algorithmRFM customer value
NETL
NSTL
万方数据
