Download PDFOpen PDF in browser

Customer Churn Prediction in Irancell Company Using Data Mining Methods

EasyChair Preprint 2422

5 pagesDate: January 20, 2020

Abstract

Churn is one of the major challenges that various businesses face in today's competitive environment. When customer becomes churn, organizations lose one of their most important assets, which can lead to financial losses and even bankruptcy. One of the major mitigation of this challenge is predicting the churn customers and with a motivational campaign try to prevent the leaving.
One of the most effective methods for predicting churned customers is the use of data mining and artificial neural Network (ANN) which are more efficient methods than other available methods in forecasting because of its diversity in structure and its different training algorithms. 
In this research, a hybrid approach based on a genetic algorithm and modular artificial neural network is presented to predict the customers that likely to be churned in telecom companies. The purpose of applying the Genetic Algorithm is to obtain the optimal structure of modules in the modular ANN so that it can provide the highest accuracy in predicting churn. Examination of the results of the proposed method and comparing it with other methods shows that the proposed method is superior to other executed methods with an accuracy of about 95.5% for US telecommunications data (which was used for assessing) and 88.1% for data set of Irancell. The reason for this result is due to the use of a genetic algorithm to providing the optimal structure for modules of ANN and also the use Feedforward Neural Network.
 

Keyphrases: Artificial Neural Network, Churn, Data Mining, Genetic Algorithm, Irancell, Telecom

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:2422,
  author    = {Homa Meghyasi and Abas Rad},
  title     = {Customer Churn Prediction in Irancell Company Using Data Mining Methods},
  howpublished = {EasyChair Preprint 2422},
  year      = {EasyChair, 2020}}
Download PDFOpen PDF in browser