Adaptable model based on ensemble learning for different telecommunication data

Abstract

The ultimate goal in designing and recommending an appropriate tariff plan is to be able to predict customers' behavioral patterns in light of the current situation in the telecommunications market. The clients' behavioral patterns and their background in terms of demographics are quite important. The study model put forth in this paper uses a variety of machine learning techniques to anticipate customers' behavioral patterns based on their demographic information. The model was developed after researching a number of classification-based machine learning techniques, including some ensemble techniques like random forest, adaboost, gradient boosting machine, extreme gradient boosting, bagging, and stacking, as well as more conventional ones like decision tree, k-nearest neighbor, logistic regression, and artificial neural networks. Understanding consumer needs is important, but the telecommunications business also needs to be able to anticipate customer attrition. The goal is to use the same research methodology to anticipate customer turnover rates more accurately while maintaining profit. With the suggested model's ability to function on many dataset types, the main goal has been accomplished.