Research on Financial Performance Evaluation of Agricultural Listed Companies Based on BP Neural Network Model

This study aims to create a financial performance evaluation model for agricultural listed companies using the BP neural network, addressing traditional systems' inaccuracies. Based on the characteristics of these companies, a financial performance evaluation indicator system is constructed, with 42 A-share listed agricultural enterprises as the sample. The model includes an input layer with 17 neurons for key financial indicators, an optimized hidden layer, and an output layer for financial performance. Trained using the Levenberg-Marquardt algorithm, the model's performance is evaluated by MSE and R-squared metrics. Results show high accuracy, with MSE near zero and R² reaching one, indicating a strong fit between predictions and actual data. This BP neural network model offers a reliable tool for assessing agricultural enterprises' financial performance, providing valuable insights for management and decision-making.