| Popis: |
The phenomenon of customer churn, which occurs when a customer leaves a service provider for another, is a major challenge for businesses in the telecommunications industry. Accurate predictions of this issue can help them improve their profitability and reduce their customer attrition. This study aims to use the data collected by Orange to improve the prediction accuracy of this issue. The study begins by reviewing the Telecom Churn dataset, which contains details about the customers such as their demographics and usage patterns. It also has a label that indicates whether or not the customer has churned. Through exploratory data gathering, we can identify correlations, patterns, and possible predictive elements that can be utilized in the prediction of churn. The goal of this study is to develop an ensemble learning framework that combines multiple classifiers. The framework is composed of various classification algorithms- Stochastic Gradient Boost(SGD), Random Forests (RF), Gradient Boosting (GB) and AdaBoost. We also test the performance of these through cross-validation techniques. The goal of this research is to improve the accuracy of its predictions by capturing more accurate information about the behavior of customers. The evaluation of the ensembles involves using different performance metrics, such as accuracy, recall, and F1 score. According to our experiments, the ensemble learning GB outperforms the other classifiers when it comes to predicting the likelihood of a customer leaving a service provider. By incorporating the base classifiers' predictions, the ensembles were able to achieve a robust and accurate prediction. This method can help businesses identify potential churners and implement effective retention strategies. The findings of this study demonstrate the utility of ensembles in improving the accuracy of churn prediction models for telecommunications companies. These findings can be utilized to develop more reliable and accurate churn prediction models, which can help improve the customer retention rate and enhance the business performance of service providers. |
