A Comparative Study of Multi-Model Prediction of Deep Learning Ability-taking Middle School Mathematics Learning as an Example
Abstract
Deep learning is increasingly being applied in educational projects. Predicting and evaluating deep learning in middle school mathematics is crucial for optimizing educational project practice. This study, based on Huberman's 5C deep learning model and the NESS-China scale, designed a questionnaire to investigate the current status of deep learning in mathematics among middle school students in Yantai, China. A prediction model for deep learning ability in mathematics was constructed using linear regression, decision tree, and neural network algorithms. The scores on each dimension of deep learning served as independent variables, and the students' overall learning ability level served as the dependent variable. Model performance was evaluated using mean squared error, mean absolute error, and coefficient of determination. The results showed that the linear regression model achieved the best predictive performance, indicating a significant linear correlation between students' deep learning ability and performance on each dimension. This finding suggests that teachers can focus on cultivating students' learning abilities in key dimensions to maximize their overall learning ability. This study provides new insights and empirical evidence for tailoring teaching to individual students and promoting deep learning in mathematics, and offers a fundamental research basis for the application of deep learning in educational projects.