Construction of College English Teaching Quality Evaluation Model Based on RBF Neural Network Algorithm

With the acceleration of globalization and the internationalization of education, the importance of College English teaching quality evaluation (CETQE) has become increasingly prominent. Traditional methods, such as questionnaire survey and students’ grading, are simple and easy to implement, but they are easily influenced by subjective factors, so it is difficult to comprehensively and objectively reflect the teaching quality. For this reason, this paper introduces RBF neural network (RBFNN), which uses its powerful nonlinear mapping ability and adaptability to simulate and capture the complex relationships in TQE. Firstly, principal Component Analysis (PCA) is employed to identify the crucial TQE indicators across various dimensions, such as instructional content, pedagogical approaches, interactions between educators and students, and educational outcomes, thereby ensuring both the thoroughness and precision of the assessment. Subsequently, a RBFNN architecture is devised, comprising an input layer, a concealed layer, and an output layer. The input layer incorporates five handpicked evaluative metrics, the hidden layer utilizes the Gaussian function as its activation mechanism, with its core vector and breadth parameters being initialized via the K-means clustering technique. Conversely, the output layer ascertains the teaching quality score through a process of linear amalgamation. Empirical findings demonstrate that the TQE model, grounded in RBFNN, markedly outperforms conventional models like linear regression and decision trees in terms of precision and recall. Within a dataset of 103 instances, the RBFNN model’s accuracy on the testing set attains 90%, while the recollection rates for superior and substandard instructional categories stand at 85% and 80% respectively. In addition, by comparing with the expert evaluation results, the identification value of RBFNN model is close to the expert evaluation score, which verifies its effectiveness and reliability. The TQE model based on RBFNN constructed in this paper not only provides TQE with more scientific and objective methods and tools, but also provides new ideas for future research in related fields.