An Improved Clustering Algorithm and Its Application in WeChat Sports Users Analysis
Xuanxia Yao, Shuying Ge, Huafeng Kong, Huansheng Ning · Procedia Computer Science · 2018
Determining the number of clusters is an important issue in clustering, which can be either designated artificially or determined automatically. For the latter, it’s critical to design an appropriate method to update clusters number. Although many researches have been made for numerical, categorical or mixed datasets, most of them are not very effective or cannot guarantee the unique clustering result. To address these problems, an improved clustering algorithm based on entropy is put forward, which uses the divergence to determine the initial cluster centers and introduce the inter-cluster entropy for mixed data to update clusters number. The experiments on the 3 dataset in UCI and the practical dataset from WeChat sports users show that the improved algorithm is a deterministic clustering algorithm with good performance.