Churn Prediction using Cuckoo Search Optimization and Enhanced Kernel Extreme Learning Machine Algorithm Over Telecom Data

  • Neeraj Varshney et al.

Abstract

Customer retention is viewed as one of the principle worries of organizations in the media communications industry. By the expanding rivalry and decent variety of contributions available, numerous media communications organizations exploit information mining methods to foresee client stir. This examination works a certifiable report on Customer churn expectation and proposes the use of classifier to improve a Customer churn prediction model. To predict churn behavior of individual clients is the significant target of this work. In this research, Cuckoo Search (CS) optimization-Enhanced kernel Extreme Learning Machine (EELM)classifieralgorithm is projected to progress the churn prediction presentation significantly. This research comprises four chiefstages such as preprocessing, customer behaviour analysis, characteristicchoice and Customer churn prediction. The preprocessing step is used to improve the missing values and remove the unnecessary values effectively. Then KNN algorithm is applied to analyze the customer behaviour using k-closest similarity values. The four conditions are considered like client disappointment (H1), exchanging costs (H2), service use (H3) and client status (H4). In characteristic determination, CS picked the best enlightening properties utilizing wellness esteems. EELM is applied in this effort as a premise forecast prototype for Customer agitate expectation. In this way the test result infers that the projected CS-EELM strategy gives improved beat forecast execution regarding higher exactness, review, f-measure, and precision.

Published
2019-12-12
How to Cite
et al., N. V. (2019). Churn Prediction using Cuckoo Search Optimization and Enhanced Kernel Extreme Learning Machine Algorithm Over Telecom Data. International Journal of Control and Automation, 12(6), 147 - 156. Retrieved from https://sersc.org/journals/index.php/IJCA/article/view/2029