2019/01/18 by Uzair Ahmed, Asifullah Khan, Ahmed, Uzair +10
Business, Management and Accounting · #Consumer Market Behavior and Pricing #Consumer Retail Behavior Studies #Customer churn and segmentation #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.1901.06091
openalex publication_date 2019/01/18 · openalex created_date 2022/07/30 · openalex updated_date 2026/07/28
A churn prediction system guides telecom service providers to reduce revenue\nloss. However, the development of a churn prediction system for a telecom\nindustry is a challenging task, mainly due to the large size of the data, high\ndimensional features, and imbalanced distribution of the data. In this paper,\nwe present a solution to the inherent problems of churn prediction, using the\nconcept of Transfer Learning (TL) and Ensemble-based Meta-Classification. The\nproposed method TL-DeepE is applied in two stages. The first stage employs TL\nby fine-tuning multiple pre-trained Deep Convolution Neural Networks (CNNs).\nTelecom datasets are normally in vector form, which is converted into 2D images\nbecause Deep CNNs have high learning capacity on images. In the second stage,\npredictions from these Deep CNNs are appended to the original feature vector\nand thus are used to build a final feature vector for the high-level Genetic\nProgramming (GP) and AdaBoost based ensemble classifier. Thus, the experiments\nare conducted using various CNNs as base classifiers and the GP-AdaBoost as a\nmeta-classifier. By using 10-fold cross-validation, the performance of the\nproposed TL-DeepE system is compared with existing techniques, for two standard\ntelecommunication datasets; Orange and Cell2cell. Performing experiments on\nOrange and Cell2cell datasets, the prediction accuracy obtained was 75.4% and\n68.2%, while the area under the curve was 0.83 and 0.74, respectively.\n