Course assessment model in vocational education based on BPNN optimized by genetic algorithm

With the current educational system reform, vocational education has become an increasingly popular learning path for Chinese students, making the optimization of course assessment models a critical issue. However, traditional evaluation systems face problems such as subjective judgment and reliance on large amounts of training data, resulting in high labor and time costs. To address these issues, this study proposes a vocational education assessment model based on a Genetic Algorithm combined with a Backpropagation Neural Network. The model processes data using normalization techniques and employs the entropy method to determine the weights of evaluation indicators. Additionally, the Genetic Algorithm is used to optimize the model’s weights and thresholds, while the Levenberg-Marquardt method is applied for data training. The experimental results show that the Root Mean Squared Error decreased from 12.2% in the training phase to 7.8% in the testing phase. After 100 iterations, the model’s loss value dropped to 0.13, and the recall rate increased from 65 to 90%. The precision of the assessment results also improved from 87.9% in the training set to 96.5% in the validation set. For the self-built validation dataset, the model’s Coefficient of Determination stabilized at 0.96 after 100 iterations, and the relative error decreased from 17.1 to 14.9%. The fitting coefficient between the predicted and actual assessment values was 0.998, with an average error of 0.67%, outperforming the comparison model. These results demonstrate that the proposed model can handle various vocational education assessment scenarios with high accuracy, providing innovative ideas for the development of vocational education.