Churn Prediction Using Machine Learning and Recommendations Plans for Telecoms

Abstract

Keeping customers satisfied is truly essential for saying that business is successful especially in the telecom. Many companies experience different techniques that can predict churn rates and help in designing effective plans for customer retention since the cost of acquiring a new customer is much higher than the cost of retaining the existing one. In this paper, three machine learning algorithms have been used to predict churn namely, Naïve Bayes, SVM and decision trees using two benchmark datasets IBM Watson dataset, which consist of 7033 observations, 21 attributes and cell2cell dataset that contains 71,047 observations and 57 attributes. The models' performance has been measured by the area under the curve (AUC) and they scored 0.82, 0.87, 0.77 respectively for IBM dataset and 0.98, 0.99, 0.98 respectively for cell2cell dataset. The proposed models also obtained better accuracy than the previous studies using the same datasets.

Impact of Hyperparameters on Deep Learning Model for Customer Churn Prediction in Telecommunication Sector

“…e sample distribution is modified in the AdaBoost algorithm, which eliminates some of the mistakes from previous iterations [24].…”

“…According to the current studies, these techniques improve the prediction performance according to current studies. For example, Ebrah and Elnasir [24] applied the algorithms (Naive Bayes, support vector machine, and decision tree) to train the dataset to predict customer churn for a telecommunication operator and found that the support vector machine algorithm had better performance. e work of Brandusoiu and Toderean [13] also confirmed the effectiveness of these techniques using logistic regression and SVM.…”

Churn prediction in telecommunication industry using kernel Support Vector Machines

“…Jain et al [ 15 ] proposed two approaches including Logistic Regression and Logit Boost but the accuracy generated from these methods were lowers than models using kernel tricks introduced in preceding studies which used the same dataset. Furthermore, in [ 3 , 5 , 7 , 10 ], authors showed that kernel Support Vectors Machines gave the best results compared to other techniques. Each real-world dataset has its own properties and kernel tricks are very functional in constructing the type of space and hyperplanes of Support Vector Machine that are suitable with the studied dataset.…”

“…Many previous works [ 1 – 10 ] have confirmed that customer churn prediction depended on complex human behaviors, such as customer data, usage, consumption behavior, and so on.…”

“…Most previous studies focused on improvements of predictive models using machine learning algorithms [ 1 – 10 ] such as neural networks, support vector machine, random forest, etc. However, prediction results are not highly accurate due to incorrect classifications of churners and non-churners.…”