Vocational education teaching quality assessment based on deep learning and multi-objective optimization decision algorithms

Abstract Vocational education is a critical component of the economy and employment; therefore, measuring and evaluating teachers’ quality is essential to improve student engagement and achievement levels. This investigation aims to build a comprehensive framework for assessing and optimizing a vocational program’s quality of teacher instruction. The research combines a Multi-Objective Decision Evaluation (MODE) established on Deep Convolutional Networks (DCN) for feature extraction and optimizing various vocational program objectives. The Teaching Quality Assessment Data Set contains demographic data, student input, faculty evaluation, and academic performance. The data was further cleaned, normalized, and scaled to ensure quality and consistency in the teaching quality assessment data set. Dimensionality reduction through principal components analysis (PCA) allows for the retention of the most significant characteristics, and recursive feature elimination (RFE) removes least important characteristics from the data set, which will provide a better-performing and interpretable model. Further, the MODE–DCN adopts the preprocessed data to extract relevant patterns and improve teaching style according to multiple criteria. The implementation of the model and its optimization were completed using TensorFlow and Python, yielding performance metrics of accuracy (98.5%), F1-score (96.6%), precision (96.3%), and recall (95.8%) on the testing dataset. The MSE was reduced to 14.67 and resulted enhanced model constancy and estimate reliability. The suggested approach proves a considerable improvement over other models to assess teaching quality and offers an approach that optimizes multiple objectives in equal measure.