Intelligent Data Analytics using Hybrid Gradient Optimization Algorithm with Machine Learning Model for Customer Churn Prediction

Elvir Akhmetshin¹, Nurulla Fayzullaev², Elena Klochko³, Denis Shakhov⁴, Valentina Lobanova⁵

¹Candidate of Economic Sciences, Associate Professor of Department of Economics and Management, Kazan Federal University, Elabuga Institute of KFU, Elabuga, Russia.

²Candidate of Economic Sciences, Associate Professor, Head of Department of Accounting and Audit, Urgench State University, Urgench, Uzbekistan.

³Doctor of Economic Sciences, Professor of Department of Management, Kuban State Agrarian University named after I.T. Trubilin, Krasnodar, Russia.

⁴Candidate of Sociological Sciences, Associate Professor of Department of Economics and Management, Khorezm University, Urgench, Uzbekistan.

⁵Candidate of Economic Sciences, Associate Professor of Department of Enterprise Economics, Regional and Personnel Management, Kuban State University, Krasnodar, Russia.

Emails: elvir@mail.ru; fayzullaev.n.b@mail.ru; klochko.e.n@yandex.ru; d.a.shkhv@gmail.com; valentina.v.lobanova@list.ru

Abstract

Intelligent data analytics for customer churn prediction (CCP) harnesses predictive modelling algorithms, machine learning (ML) techniques, and advanced big data analytics and also uncovers the underlying drivers and patterns of churn and detects customers at risk of churning. This business strategy help organization to implement retention efforts to decrease customer attrition and proactively detect at-risk customers. CCP allows businesses to take proactive measures such as targeted marketing campaigns, personalized offers, or enhanced customer service, to maintain valuable customer and decrease revenue loss. It is widely used in industries like telecommunications, subscription services, e-commerce, and finance to optimize customer retention strategies and enhance long-term profitability. ML algorithm can detect indicator and underlying trends that precedes churn by analyzing historical customer data, including transactional patterns, behaviors, demographics, and customer interaction. The study introduces Intelligent Data Analytics using Hybrid Gradient Optimization Algorithm with Machine Learning (IDA-HGOAML) Model for Customer Churn Prediction. The main intention of IDA-HGOAML method focuses on the prediction and classification of customer churns and non-churns. To do so, the IDA-HGOAML technique initially undergoes data pre-processing using Z-score normalization. The IDA-HGOAML model makes use of equilibrium optimization algorithm (EOA) for the feature selection (FS). Besides, the churn prediction method is implemented by the convolutional autoencoder (CAE) model. Finally, the HGOA is exploited for the optimal hyperparameter selection of CAE model, thereby enhancing the prediction results. A widespread experimental analysis were performed to validate the enhanced efficiency of the IDA-HGOAML method. The extensive outcomes indicated the improved prediction results of the IDA-HGOAML method over existing techniques in terms of different measures.

Keywords: Intelligent systems; Data analytics; Customer churn prediction; Feature selection; Machine learning