Financial Products; Recommendation Systems; Fintech; Classification.

Recommendation systems play a crucial role in various sectors, including finance, by providing personalized suggestions to users based on their story and preferences. In the context of the financial industry, these systems have the potential to revolutionize how institutions interact with customers, especially by tailoring recommendations for financial products such as investments, insurance, and loans. In this scenario, institutions can increase customer satisfaction and loyalty while improving their returns at the same time. The aim of this study was to develop a classification model capable of predicting customer tendency to subscribe to a insurance service, an then use it as a recommendation system in a fintech. Initially, exploratory data analysis was conducted, followed by preprocessing techniques such as feature removal to prevent data leakage and normalization. The resulting dataset showed class imbalance. To establish a baseline, a decision tree classifier was applied in conjunction with grid search for hyperparameter optimization in three distinct experiments. In the final experiment, the dataset was balanced to enhance the model’s predictive capacity. The preliminary results for the model with optimized parameters were accuracy of 58.95%, precision of 58.28%, recall of 63.70%, and F1-score of 60.87%, showing promising performance for the initial baseline model.