Despite the increasing application of machine learning in loan allocation, limited research has explored its potential to enhance supply-side decision-making in personal loan marketing. This study investigates the predictive effectiveness of four machine learning algorithmsdecision tree, random forest, k-nearest neighbours, and logistic regressionin forecasting consumer acceptance of personal loan offers. It also provides practical recommendations for neo banks and other financial institutions seeking alternatives to traditional segmentation methods. Utilising a real-world dataset comprising 5,000 customer records from Neo Bank, each model was evaluated using classification metrics including accuracy, precision, recall, and F1-score. The findings indicate that the pre-pruning decision tree model outperforms the others, achieving the highest performance (accuracy: 0.9860; recall: 0.9329; precision: 0.9267; F1-score: 0.9298). Additionally, income, family size, and education level were identified as key predictors of loan acceptance. By leveraging precision marketing, customer-centric loan product design, and continuous model optimisation, Neo Bank can significantly improve personal loan conversion rates, reinforce a customer-first brand image in a competitive financial environment, and drive long-term sustainable growth.
Lyu et al. (Wed,) studied this question.