Optimization of credit marketing strategy based on customer demand and RNN

Optimizing credit marketing strategies not only improves banks’ economic performance but also meets the evolving needs of the market economy. To achieve this goal, this paper puts forward an integrated optimization strategy from two perspectives: credit usage prediction and credit interval prediction. For credit usage prediction, the paper applies decision trees to extract business rules and filters them accordingly. Meanwhile, a credit usage prediction model is built using Deep Neural Networks, where Recurrent Neural Networks are used to capture time-series features and Categorical Boosting models are used to process non-time-series features. Experimental results show that the credit usage prediction model achieves an accuracy of 95.2% to 97.5% and an Area Under the Receiver Operating Characteristic Curve ranging from 0.93 to 0.95 on two datasets, significantly outperforming baseline models. The credit interval prediction model reaches a maximum Mean Absolute Error of 3.651, a minimum Mean Squared Error of 23.250, and a coefficient of determination greater than 0.896. In addition, the integrated strategy based on both models shows strong adaptability across different scenarios. These findings suggest that the proposed models demonstrate reliable predictive performance and can provide analytical support for optimizing credit marketing strategies.