A Termination Prediction for Postpaid Mobile Service using Machine Learning: The Case of Ethio Telecom
| dc.contributor.advisor | Ephrem, Teshale (PhD) | |
| dc.contributor.author | Tewabe, Kassaw | |
| dc.date.accessioned | 2022-02-11T04:41:39Z | |
| dc.date.accessioned | 2023-11-04T15:13:04Z | |
| dc.date.available | 2022-02-11T04:41:39Z | |
| dc.date.available | 2023-11-04T15:13:04Z | |
| dc.date.issued | 2022-02 | |
| dc.description.abstract | Customer churn is a major issue for operators that may greatly affect their revenue. Even if in Ethiopia’s monopolistic market, there is still a challenge in the form of subscriber service termination. To solve this issue, operators must first identify potential service termination in advance and then take proactive measures to reduce the number of terminations. For this purpose, telecom operators need a prediction model to predict correctly and timely potential service termination subscribers based on collected service usage-related data. To anticipate subscriber behavior, the performance of several prediction methods has been explored. However, the performance of such algorithms is not examined in the context of Ethiopia postpaid mobile service and in the case of multi-class, where ethio telecom possesses multi-class data that enable develop a multi-class prediction model for mobile subscriber service termination. The goal of this study is to investigate the performance of prediction learning algorithms with multi-class scenarios for predicting service termination in the context of Ethiopian postpaid mobile subscribers. The algorithms investigated include J48 Decision tree, Random Forest (RF), and Multilayer Perceptron (MLP), and their performance is measured in terms of prediction accuracy, precision, recall, and F-Measure. Cross-validation (k=10) techniques and a multi-class dataset are used to test the performance of algorithms. WEKA 3.9.4 tool algorithm implementation was utilized for performance evaluation. As a result, the J48 and RF prediction algorithms almost have the same performance on all performance parameters result. However, MLP algorithm achieved a lowperformance score compared to J48 and RF, and the accuracy of J48, RF, and MLP are 94.9%, 95.1%, and 93.3% respectively. | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/123456789/30012 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | Postpaid mobile service | en_US |
| dc.subject | Termination prediction | en_US |
| dc.subject | Machine learning | en_US |
| dc.subject | Multi-class | en_US |
| dc.title | A Termination Prediction for Postpaid Mobile Service using Machine Learning: The Case of Ethio Telecom | en_US |
| dc.type | Thesis | en_US |