Handling Class Imbalance in Customer Churn Prediction in Telecom Sector Using Sampling Techniques, Bagging and Boosting Trees

Introduction: The Churn means Attrition. In the world of competition and customer needs, the superior quality service, to predict customer churn, has become a pressing issue for the HRMS Vendors or service providers. Machine Learning is capable to solve this challenging problem. The customer churn can be predicted, and proactively, HRMS vendors can apply a strategy to retain customers and manage the customer relationship. Objectives: The objective of this paper is to meticulously analyse and summarize the previous studies conducted for customer churn prediction. Furthermore, this paper will serve as a valuable resource for future research in the context of HRMS vendors or service providers. Methods: This paper focuses on machine learning algorithms like Decision Tree, Random Forest, Support Vector Machine, Gradient Boosting, Neural Network, etc. The paper highlights the accuracy of the outperformed models. This paper includes the previous studies between 2021 and 2025. In total, 30 papers were reviewed meticulously. To provide more clarity, all the previous studies are summarized in a table format. The paper was searched on Google with the keywords ‘Customer Churn’ AND ‘Machine Learning’. The Duplicates were removed, and the previous studies were analyzed. Results: The findings showcase that various machine learning models like RF, DT, Logistic Regression, Gradient Boosting are applied to predict churn. Furthermore, by this analysis it is clearly stated that hybrid or ensemble models work best for customer churn model prediction. This study guides on selecting the best model as per the relevant industry. Conclusions: This comprehensive survey has meticulously analysed the implementation of machine learning algorithms to predict customer churn and highlighted its importance to fully transform the business to build client retention strategies. In this paper, different machine models are examined, and for each previous study, the machine learning models used & their results are discussed.

Customer churn is a critical concern for the telecommunication industry. Understanding and predicting customer churn can lead to more effective retention strategies and an increase in profitability. Predicting customer churn allows telecommunication companies to identify potentially dissatisfied customers early on and take proactive measures to retain them. Due to a large client base, the telecom industry generates a large volume of data on a daily basis. Decision makers and business analysts stressed that acquiring new customers is more expensive than retaining existing ones. Business analysts and customer relationship management (CRM) analysts must understand the reasons for customer churn as well as behaviour patterns from existing churn data. This paper proposes a churn prediction model that uses classication and clustering techniques to identify churn customers and provides the factors that contribute to customer churning in the telecom sector. The results presented shows that XBoost and Random Forest achieved higher prediction accuracy when compared to K- Nearest Neighbours, Support Vector Machines and Decision Trees in terms of accuracy, precision, F1-Score and recall.

This thesis proposes customer churn prediction model for a Software-as-a-Service inventory management company in Thailand. Software-as-a-Service is the fast growing and high market values industry as a new emerging online business. Customer churn is a critical measure for this business. Thus, this thesis focuses on seeking a customer churn prediction model specified for a Software-as-a-Service inventory management company in Thailand which is facing a high churn rate issue. This thesis executes the prediction models with four machine learning algorithms: logistic regression, support vector machine, decision tree and random forest. The results show that the optimized decision tree model is capable to outperform other classification models toward recall scorer with validated testing scores of 94.4% of recall and 88.2% of F1-score. Moreover, feature importance scores can highlight useful insights of case-study that business metrics are significantly related to churn behavior. As a result, this paper is beneficial to the case-study company to help indicate real churn customer correctly and enhance the effectiveness in making executive decision and marketing campaign.

The telecommunication industry need a customer churn prediction due to many competitors. The companies also lack of churn prediction to retain the customer. This problem not only affect the growth of the business but also affect the revenues. To retain the existing customer is very crucial task for the company. A rapid increasing in technology, the various machine learning and deep learning are tools are developed which can be used by telecoms companies to monitor the churn behaviour of customers. In this study, a brief idea on the customer churn problem on various machine learning techniques such as XGBoost, Gradient Boost, AdaBoost, ANN, Logistic Regression and Random Forest are analysed. Also the various deep learning techniques such as Convolutional Neural Network, stacked auto encoders to predict the customer churn problem are analysed by comparing the models in terms of accuracy.

In today's conditions, customer loyalty has gained importance with the increase in the competitive environment between companies, the development of marketing strategies and the improvement of companies. Therefore, it is essential to acquire customers for a company to survive. Retaining an existing customer in the telecommunication sector is less costly than gaining a new customer. Customer churn analysis is the process of predicting customers with high abandonment requests by examining the offers and utilizable behaviors. Customer churn analysis provides services to develop various campaigns aiming to increase the company’s loyalty by predicting the customers who are planning to move to another company. In this way, it gives the company a competitive advantage. This study aims to make predictions by developing models for customer churns through data mining and machine learning methods in the telecommunication sector. In addition, we believe that the application in this article will contribute to data analysts and academicians who will want to analyze customer churn with different data sets in telecommunication and other sectors in the future. The analysis in this study is carried out on a data set obtained from an open-access database, including 20 transaction records for the customer from 7043 customers and whether the customer left the company. Among the data mining methods, Random Forest (RF), Support Vector Machines (SVM) and Multilayer Artificial Neural Networks (ANN) are modeled in open-source Phyton environment. The results have shown that ANN has fared better at classifying customers than other machine learning methods.

The telecommunication sector becomes popular in recent years and service companies suffer from a loss of valued customers to competitors. In the operational area, customer churn is a challenging issue, which makes companies, loses customers, and therefore is of bigger issue for the industry. Recent technologies and competitors were developing quickly and churn prediction becomes essential for telecom companies. Customer Churn Prediction (CCP) models offer precise detection of significant churners and therefore a retaining solution can be offered to them. Data analytics and machine learning (ML) models can be designed and used for churn investigation to reduce customer churns and improve profit. In this aspect, this article investigates a novel Tomtit Flock Optimization with Deep Variational Autoencoder for Customer Churn Prediction (TFODVAE-CCP) in the telecommunication sector. The presented TFODVAE-CCP technique mainly concentrates on the identification and classification of customer churns. To accomplish this, the presented TFODVAE-CCP technique performs data pre-processing at the primary stage. Next, the TFODVAE-CCP technique employs TFO algorithm to elect an optimal subset of features and thereby improve classification accuracy. Finally, the DVAE method can be leveraged for classifying churner/non-churner. A detailed set of experimental analyses is made on telecommunication churn dataset and the results illustrate the betterment of the TFODVAE-CCP method over other existing models.

Customer churn, a critical metric in the telecommunications industry, reflects the propensity of customers to discontinue service usage, thereby affecting the revenue and growth of a company. Customer churn is greatly affected by individuals and products, and there is no obvious rule, which greatly increases the difficulty of customer loss prediction. With the development of machine learning, the prediction in telecom churn using machine learning has attracted wide attention from researchers. However, there are great differences between different algorithms, which increases the difficulty of application. Therefore, this paper studies the issue of customer churn prediction from the aspect of algorithm. This paper introduces the common data sets and evaluation indexes of customer churn prediction, and then introduces logistic regression, decision tree, K-nearest neighbor, random forest, Gaussian naive Bayes, XGBoost, gradient enhancement and neural network algorithms. Then, through the experimental analysis, the differences between these models are compared. Finally, the paper summarizes the full text and discusses the future research hotspots.

Anti-tuberculosis drug-induced liver injury (ATLI) is the most harmful to anti-tuberculosis (TB) treatment. This study aims to construct and validate a binary ATLI risk prediction model based on clinical data through seven machine learning algorithms (logistic regression, decision tree, support vector machine, random forest, gradient boosting decision tree, extreme gradient boosting, and light gradient boosting machine [LightGBM]). A retrospective cohort of 2356 TB patients followed between January 2017 and December 2024 was used to develop and evaluate the prediction model. Random undersampling was performed to address class imbalance problem. Least absolute shrinkage and selection operator (LASSO) regression was used to select features and retained 27 of 31 original features. Seven ML algorithms were trained and the LightGBM model demonstrated optimal performance in testing set based on the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity (AUC = 0.789, sensitivity = 0.734, and specificity = 0.706). The model exhibited optimal simplicity and stability when incorporating the 8-feature combination comprising baseline high-density lipoprotein cholesterol (HDLC), γ-glutamyl transpeptidase (GGT), triglycerides, total cholesterol (TCHOL), uric acid, total bilirubin (TBIL), globulin (GLB), and liver disease history (AUC = 0.764, sensitivity = 0.758, and specificity = 0.610), and Shapely additive explanations analysis also revealed that these variables were the most influential contributors. The optimal model maintained robust predictive ability in the external validation cohort (AUC = 0.721, sensitivity = 0.828, and specificity = 0.604). This study determined that the combination of baseline HDLC, GGT, triglycerides, total cholesterol (TCHOL), uric acid, TBIL, GLB, and liver disease history was an important predictor of ATLI through LightGBM model, which could help clinicians in the early identification of ATLI.

The classification model in machine learning has been employed to address different problems. Machine Learning classification is an effective method to realize customer churn prediction. This article provides a comparative study of machine learning from the perspective of predicting customer churn. Customer churn is one of the obstacles hindering the development of companies. Through classification approaches based on machine learning, customer churn can be predicted precisely, thus providing decision-making capabilities to these companies. Customer churn prediction is a typical classification problem and can be addressed by the use of a decision tree or random forest model. In this paper, decision tree and random forest models are employed to predict customer churn, using sales data provided by a chemical company from 2012 to 2020. We analyze the underlying risk of customer churn and the customer churn factors, including the low-priced count (LC), total amount of money (TM), and creation time (CT). The prediction results of the two models are evaluated by calculating various metrics. The experimental results indicate that the low-priced count (LC) is the most essential factor for customer churn. The results of the comparison, considering the training error and generalization error including: confusion matrix, ROC curve, AUC, precision, recall, and F1 score, reveal that the random forest model has better prediction accuracy than the decision tree model.

Customer churn prediction plays a vital role in modern business, accurately influencing strategic and operational decisions that influence customer loyalty to a service. Customer churn focuses on customer retention being more profitable than attracting new customers because long-term customers provide lower profits and costs while losing customers increases the costs and need to attract new customers. However, customer churn still occurs frequently and cannot be predicted. If customer churn is left unchecked, it will endanger the company or banking industry because it can cause loss of income, damage reputation, and decrease market share. Random Forest, a data mining technique, was used in this research because of its ability to predict and handle many variables. This research aims to predict customer churn using the Random Forest method with datasets from Europe, especially France, Spain, and Germany, hoping to benefit the banking industry by identifying customers at high risk of abandoning services. This research is expected to benefit business people from customer churn predictions. Especially in the banking industry, it can help identify customers at high risk of abandoning service. Thus, companies can take appropriate steps to retain these customers, increase customer retention, strengthen customer loyalty and optimize their business performance. The results of this research are an accurate system for predicting customer churn in the future. The research obtained accuracy results of 87% in predicting customer churn using accuracy testing in the form of a confusion matrix.

Alzheimer’s disease (AD) is a gradient degeneration of essential cognitive activities such as memory, thinking, and cognition. AD mainly affects elderly individuals and is recognized as the most common cause of dementia. This study investigates the predictive performance of nine supervised machine learning algorithms—Logistic Regression, Decision Tree, Random Forest, K-Nearest Neighbors, Support Vector Machine, Gaussian Naïve Bayes, Multi-Layer Perceptron, eXtreme Gradient Boost, and Gradient Boosting—using neuropsychological assessment data. We applied two classification techniques—binary and multiclass—to classify 1761 subjects into three categories: cognitively normal (CN), mild cognitive impairment (MCI), and Alzheimer's disease (AD). Binary classification tasks focused on CNvsAD and CNvsMCI subsets, while multiclass classification used the full dataset (TriClass). Hyperparameter tuning was performed to optimize model performance. The results indicate that ensemble learning models, particularly Gradient Boosting (GB) and Random Forest (RF), exhibited superior accuracy compared to other algorithms. Most models for the CNvsAD subset achieved the highest accuracy (97.74%), while GB achieved the best performance (94.98%) for the CNvsMCI subset. For multiclass classification, RF achieved the highest accuracy at 84.70%. These findings highlight the robustness and efficiency of ensemble learning algorithms, especially in handling complex, non-linear data structures. This study underscores the potential of RF and GB as reliable tools for early detection and classification of Alzheimer’s disease using neuropsychological data.

Discovery of new antibiotics is the need of the hour to treat infectious diseases. An ever-increasing repertoire of multidrug-resistant pathogens poses an imminent threat to human lives across the globe. However, the low success rate of the existing approaches and technologies for antibiotic discovery remains a major bottleneck. In silico methods like machine learning (ML) deem more promising to meet the above challenges compared with the conventional experimental approaches. The goal of this study was to create ML models that may be used to successfully predict new antimicrobial compounds. In this article, we employed eight different ML algorithms namely, extreme gradient boosting, random forest, gradient boosting classifier, deep neural network, support vector machine, multilayer perceptron, decision tree, and logistic regression. These models were trained using a dataset comprising 312 antibiotic drugs and a negative set of 936 non-antibiotic drugs in a five-fold cross validation approach. The top four ML classifiers (extreme gradient boosting, random forest, gradient boosting classifier and deep neural network) were able to achieve an accuracy of 80 per cent and above during the evaluation of testing and blind datasets. We aggregated the top performing four models through a soft-voting technique to develop an ensemble-based ML method and incorporated it into a freely accessible online prediction server named ABDpred ( http://clinicalmedicinessd.com.in/abdpred/ ).

Heart disease is among the conditions that people suffer from most frequently. Millions of people worldwide pass away each year as a result of it, making it one of the main causes of mortality. Heart disease can be characterised by issues with the heart valves, heart failure, arrhythmias, and coronary artery disease. Heart disease comes in more than 30 distinct forms. By allowing for prompt intervention and the right kind of care, early and precise cardiac disease prediction can greatly improve patient outcomes. In this model, we investigate the application of machine learning techniques for anticipating cardiac disease. We investigate a large dataset made up of patient details, such as demographics, medical histories, and clinical measures. It is absolutely mind-blowing to think that machine learning algorithms could one day properly forecast when cardiac disease will start and how to diagnose it. Machine learning techniques including logistic regression, decision trees, XGBoost, gradient boosting, random forests, support vector machines (SVMs), and artificial neural networks (ANNs) are utilised to construct predictive models. A hybrid model including ANN, gradient boost, Decision Tree, SVM, random forests, & Logistic Regression makes up the forecasting model. To increase the model's accuracy. To manage missing values, normalise features and solve class imbalance. The dataset has been pre-processed. The best accurate heart disease predictions are found using feature selection approaches. Area under the receiver's operating characteristic curve, recall, accuracy, and precision are some of the performance measures used for training and evaluating the models. The major goal of this model is to put out a novel strategy for creating a model that successfully solves practical issues

Gender classification is one of the major challenges in the domain of speech analysis now-a-days. Identification of gender from acoustic properties of voice such as mean, median, and frequency, is of high importance. Machine learning is used to solve this problem because it gives promising results for classification techniques. There are several algorithms that can be used to predict the gender using acoustic properties. In this work, classifiers are evaluated using 6 different machine learning algorithms. These algorithms include K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM), Support Vector Machine Using Poly kernel and Gradient Boosting (GB). For the same voice sample dataset, the accuracy, recall, precision and F1 Score are computed for all the above-mentioned classifiers. The accuracy obtained from KNN is 96.71, DT is 93.33%, Random Forest is 97.72%, SVM is 98.48%, SVM using Poly Kernel is 97.72%, Gradient Boosting is 96.96%. The predicted values of the considered evaluation metrices on test data for all above classifiers evidenced the supremacy of SVM classifier as it is best fitted model for gender identification of acoustic. SVM attained the accuracy of 98.48% which is the highest among all.

Emotion recognition plays a pivotal role in advancing Human-Computer Interaction (HCI) by enabling systems to understand and respond to human emotional states. Among various modalities, Speech Emotion Recognition (SER) stands out as a non-invasive, cost-effective, and temporally efficient approach for detecting emotions. This study leverages the RAVDESS dataset, which provides high-quality audio recordings for emotion classification. The proposed methodology involves preprocessing audio signals to remove noise, extracting temporal and spectral features in both time and frequency domains, and implementing machine learning models for multi-class emotion classification. The study evaluates and compares the performance of several classification models, including Random Forest (RF), Multilayer Perceptron (MLP), Support Vector Machine (SVM), Convolutional Neural Networks (CNN), and Decision Trees (DT). Experimental results demonstrate promising accuracy levels, highlighting the potential of these machine learning techniques in SER. This research contributes to the integration of SER within Brain-Computer Interfaces (BCI) and other emotion-aware applications, paving the way for enhanced interactive systems. Keywords – Speech Emotion Recognition -Machine Learning - Random Forest -Multilayer Perceptron - Support Vector Machine - Convolutional Neural Networks - Decision Tree - RAVDESS Dataset