| Abstract [eng] |
This paper presents functional data methods application to telecommunication data clustering. Data clustering is one of the tools for telecommunication companies to identify their clients’ behavior by using innovative machine learning methods. There are various papers on customers of telecommunication companies segmentation. For the most part, clustering is done by assuming that time is discrete. However, in recent years there has been an increasing interest in segmentation for functional data which can outperform analysis based on discrete data. Clients of telecom companies can be segmented based on their monthly usage of data, voice calls, and SMS data. Such an analysis can be applied to find the most popular usage combinations of GBs, voice calls, and SMS, which can help to create telecom offerings to customers based on their actual service usage. Clustering using functional data methods can help understand clients' behavior and apply better marketing strategies by adjusting current offers or suggesting new ones based on actual usage. |
