Abstract:

Abstract. Telecommunications is an important tool for the business world. This sector has a strong effect on our economic activities. The telecommunications sector represents 5.9% of Romania's GDP (in 2019). The telecommunications industry is a dynamic one, constantly developing and changing on a stable market. Thus, in order to reduce and prevent the loss of customers, most of the top companies in this sector use performance analysis that treats the churn behavior as a central topic. The purpose of this paper is to identify through an analytical method easy to visualize and interpret, the peculiarities and defining characteristics of churn clients and non-churn clients within the postpaid subscription segment. The analysis was performed on a data set of a large telecommunications company in Romania. To determine the profiles of the two types of clients we used an EDA (Exploratory Data Analysis) technique, called Violin Plot. In our study, this method identified the non-churn client's situation as defined by the lower values ​​of the invoice and an additional cost as close to zero. A specific feature of the churn class is related to the number of national minutes received outside the network and the sum of the minutes received.