Repozytorium publikacji - Politechnika Gdańska

Ustawienia strony

english
Repozytorium publikacji
Politechniki Gdańskiej

Treść strony

Artificial Neural Network in Forecasting the Churn Phenomena Among Costumers of IT and Power Supply Services

This paper presents an attempt to use an artificial neural network to investigate the churn phenomenon among the customers of a telecommunications operator. An attempt was made to create a data model based on the customer lifetime value (CLV) rather than on activity alone. A multilayered artificial neural network was used for the experiments. The results yielded a 99% successful identification rate for customers in no danger of leaving, while only 57% of those identified as in danger of leaving actually did so and stopped using the company's services.

Autorzy