Optimizing decision-making in financial services through machine learning: Retention, Investment, and Inclusion Perspectives

In the rapidly evolving landscape of financial services, decision-making processes have increasingly leaned on the power of machine learning (ML) to enhance predictive capabilities and strategic outcomes. This research examines the application of ML across three critical dimensions of financial decision-making: customer retention in fintech, investment sentiment analysis in banking, and financial inclusion via microfinance. Through a comparative synthesis of existing peer-reviewed studies, this paper evaluates the effectiveness of various ML algorithms—including Random Forest, Gradient Boosting, Support Vector Machines, LSTM, BERT, and Light GBM—in solving classification and regression problems relevant to financial services. Results demonstrate that ML models significantly contribute to predictive accuracy in customer churn detection, investor sentiment forecasting, and credit scoring for underbanked populations. The use of BERT in sentiment analysis outperformed traditional models in both accuracy and investment correlation, while Gradient Boosting and Random Forest consistently yielded top performance in retention and microfinance analytics. The paper also explores the ethical implications and interpretability challenges inherent in deploying ML models across sensitive financial domains. By offering a cross-functional assessment, this study aims to inform practitioners, data scientists, and policymakers about the strategic value of ML in optimizing decision-making across diverse financial contexts. Recommendations for integrating interpretable, high-performing models in real-world systems are also presented to support future development in the field.