Kazi Ekramul Hoque

Customer churn prediction is a critical issue that streaming services face as retaining existing subscribers is vital to the success of the business. Creating reliable churn prediction models is important because the costs of acquiring new customers are usually higher than those involved in retaining existing ones. In this study, we propose a big data-driven hybrid model combining a deep neural network with a machine-learning model to efficiently forecast customer churn. Our proposed model uses Long Short-Term Memory (LSTM) with a Gated Recurrent Unit (GRU) to capture the trends in subscribers’ usage patterns over time. In addition, light gradient boosting (Light GBM) is used to leverage insights from sequential modeling along with original attributes to forecast churn. Moreover, feature selection techniques like Chi-squared testing and Sequential Feature Selection (SFS) are utilized to choose the optimum set of features for our proposed model. Furthermore, several individual models, including deep learning and traditional machine learning algorithms are also evaluated and compared with our proposed hybrid model. Additionally, the study illustrates model interpretations using Shapley Additive Explanations (SHAP) and Explainable Boosting Machine (EBM) which are used for identifying influential features in streaming services enhancing customer retention efforts. These techniques provide transparency into our proposed model’s forecasting, making them more actionable and understandable for decision-makers. Extensive experimental evaluation demonstrates the hybrid model achieves best-in-class performance with 95.60% AUC and 90.09% F1 score.

Stock price forecasting has been reported as a challenging task in the scientific and financial communities due to stock prices’ nonlinear and dynamic nature. Machine learning models exhibit capabilities that allow them to handle nonlinear data and be candidate tools for stock price forecasting. In this study, an empirical evaluation of eight conventional machine learning models’ is conducted to forecast the stock price of eleven companies belonging to the Saudi Stock Exchange. Moreover, the optimal configuration of hyperparameters in each machine learning model is identified. Forecasting performance is evaluated by two well-known error metrics: Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Wilcoxson effect size is utilized to determine the impact of hyperparameter tuning by comparing tuned and un-tuned machine learning models’ forecasting performance. Empirical results indicate there are varying impacts of hyperparameter tuning of machine learning models in forecasting stock price. After tuning the hyperparameters, Support Vector Regression outperforms other forecasting models with a significant statistical difference. In contrast, Kernel Ridge Regression shows noteworthy forecasting performance without hyperparameter tuning with respect to other un-tuned forecasting models. However, Decision Tree and K-Nearest Neighbour are the poor-performing models which demonstrate inadequate forecasting performance even after hyperparameter tuning.