Predicting Termination Of Mobile Subscribers Using Classification Algorithms

Service termination is one key challenge of operators that can significantly reduce theirrnrevenue. It is still a challenge even in Ethiopian monopoly market even if users do notrnhave alternative service provider. For instance, the market has seen 1.6 percent overallrnprepaid mobile subscriber termination rate within a quarter, let alone specific servicerntermination. To address this challenge, operators need to understand causes ofrntermination and take timely proactive actions to mitigate number of terminations. Forrnthis purpose, they need to accurately and timely predict subscribers with potential servicerntermination based on collected service related data. Performance of various classificationrnalgorithms have been investigated to predict subscribers’ behavior. However,rnperformance of such algorithms are not studied in the context of Ethiopia where ethiorntelecom has an enormous data that helps to define its subscribers’ behavior. rnObjective of this thesis work is to study performance of classification learning algorithmsrnfor predicting termination in the context of Ethiopian prepaid subscribers. Studiedrnalgorithms are J48, Random Forest and Naïve Bayes and their performance are comparedrnin terms of prediction accuracy, performance, errors and interpretability of their model.rnAlgorithms effectiveness and efficiency are evaluated considering two validationrnmethods (Percentage Split and Cross Validation) and three data sets. For the performancernevaluation, we used WEKA 3.8 tool algorithm implementation. rnObtained results show that Random Forest scores the highest prediction accuracy whilernNaïve Bayes scores the least. Random Forest and Naïve Bayes scores their best at 93.4% andrn85.9% respectively, besides J48 scores 93.3%. J48 is as accurate & robust as Random Forest.rnMoreover, it provides the most interpretable and clear model.

Get Full Work

Report copyright infringement or plagiarism