The application of improved backpropagation neural network in college student achievement prediction

Educational institutions generate a large amount of digital data in their daily operations, which is stored in servers, forming a substantial educational data set. Extracting valuable information through practical data analysis has become a critical problem that needs to be solved urgently. Students' examination results are an essential basis for evaluating their learning status, which reflects the effect of school education to some extent. Therefore, we propose a model based on the BP network and Pandas to construct a prediction model for Pandas' performance in the first year and their successful graduation to explore the potential relationship between Pandas' performance in the freshman year and graduation, thus realizing the principle of early guidance and improvement of teaching quality. Through the random prediction experiment of 9,424 scores data of 304 students in 2017 and 2018 majoring in network engineering at a university, the accuracy rate is 96.71% after the experimental data analysis and verification, which has proved that there is a potential correlation between the students' first-year course scores and graduation. Meanwhile, the improved BP network proposed in the present research exhibits reasonable practicability and extensibility in the college student achievement prediction model. (c) 2024 by the authors; licensee Growing Science, Canada .