Open AccessPrediction of Telecom Churn using Comparative Analysis of Three Classifiers of Artificial Neural Network
Author(s) -
Youngkeun Choi,
Jae Won Choi
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.j7339.0891020
Subject(s) - computer science , artificial neural network , multilayer perceptron , artificial intelligence , machine learning , random subspace method , ibm , data mining , deviance (statistics) , oracle , analytics , classifier (uml) , revenue , materials science , software engineering , accounting , business , nanotechnology
The purpose of this study is to evaluate existing individual neural network-based classifiers to compare performance measurements to improve the accuracy of deviance predictions. The data sets used in this white paper are related to communication deviance and are available to IBM Watson Analytics in the IBM community. This study uses three classifiers from ANN and a split validation operator from one data set to predict the departure of communications services. Apply different classification techniques to different classifiers to achieve the following accuracy with 75.63% for deep running, 77.63% for perceptron, and 77.95% for autoMLP. With a limited set of features, including the information of customer, this study compares ANN's classifiers to derive the best performance model. In particular, the study shows that telecom service companies with practical implications to manage potential departures and improve revenue.