Evaluating Machine Learning Techniques for Telecom Customer Churn Prediction
Fuente:
Zenodo
Salvato in:
| Autore principale: | |
|---|---|
| Natura: | Recurso digital |
| Lingua: | inglese |
| Pubblicazione: |
Zenodo
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866902131800276992 |
|---|---|
| author | Dadakidis, Giorgos |
| author_facet | Dadakidis, Giorgos |
| contents | <p>Customer churn is a persistent and costly challenge for companies in the <br>telecommunications sector, where maintaining existing subscribers is often more <br>profitable than acquiring new ones. Accurately identifying customers who are likely to <br>leave is critical for enabling targeted retention strategies. However, churn prediction is <br>complicated by significant class imbalance, as the number of churners typically <br>represents a small fraction of the overall customer base.</p> <p>This thesis explores the application of machine learning techniques to the churn <br>prediction problem using a structured experimental approach. Five experimental settings <br>were designed to evaluate and improve model performance under imbalanced data <br>conditions: a baseline scenario using the original dataset, a cost-sensitive learning setup <br>with class weighting, a recall-optimized configuration through hyperparameter tuning, an <br>experiment incorporating synthetic oversampling (SMOTE) and a final experiment using <br>the top 20 important features . A variety of classification models were assessed, including <br>both traditional machine learning algorithms and neural networks. </p> <p><br>The study aims to investigate how different learning strategies and evaluation criteria <br>affect model behavior and performance in the context of churn prediction. Emphasis is <br>placed on addressing the imbalance issue, optimizing recall of the minority class, and <br>comparing the effectiveness of algorithmic and data-driven solutions. The findings provide <br>insights into the trade-offs and considerations involved in developing fair and practical <br>predictive models for real-world customer churn scenarios. </p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_15978511 |
| institution | Zenodo |
| language | eng |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Evaluating Machine Learning Techniques for Telecom Customer Churn Prediction Dadakidis, Giorgos churn Neural Networks, Computer Machine learning Telecommunications binary classification SMOTE Class imbalance predictive analysis <p>Customer churn is a persistent and costly challenge for companies in the <br>telecommunications sector, where maintaining existing subscribers is often more <br>profitable than acquiring new ones. Accurately identifying customers who are likely to <br>leave is critical for enabling targeted retention strategies. However, churn prediction is <br>complicated by significant class imbalance, as the number of churners typically <br>represents a small fraction of the overall customer base.</p> <p>This thesis explores the application of machine learning techniques to the churn <br>prediction problem using a structured experimental approach. Five experimental settings <br>were designed to evaluate and improve model performance under imbalanced data <br>conditions: a baseline scenario using the original dataset, a cost-sensitive learning setup <br>with class weighting, a recall-optimized configuration through hyperparameter tuning, an <br>experiment incorporating synthetic oversampling (SMOTE) and a final experiment using <br>the top 20 important features . A variety of classification models were assessed, including <br>both traditional machine learning algorithms and neural networks. </p> <p><br>The study aims to investigate how different learning strategies and evaluation criteria <br>affect model behavior and performance in the context of churn prediction. Emphasis is <br>placed on addressing the imbalance issue, optimizing recall of the minority class, and <br>comparing the effectiveness of algorithmic and data-driven solutions. The findings provide <br>insights into the trade-offs and considerations involved in developing fair and practical <br>predictive models for real-world customer churn scenarios. </p> |
| title | Evaluating Machine Learning Techniques for Telecom Customer Churn Prediction |
| topic | churn Neural Networks, Computer Machine learning Telecommunications binary classification SMOTE Class imbalance predictive analysis |
| url | https://doi.org/10.5281/zenodo.15978511 |