Optimized attention-based cascaded shuffle long-term dependent network based performance analysis of adaptive e-learning among IT professionals

The increasing need for adaptive e-learning systems among information technology (IT) professionals highlights the need for intelligent performance analysis framework for handling complex and high-dimensional data. Thus, this research introduced a deep learning based model for analyzing the performance of IT professionals in adaptive e-learning scenario. Initially, the data acquisition is employed based on behavioral, interactional and assessment metrics. The gathered data are preprocessed using noise elimination based on redundant data removal, K-nearest neighbor based imputation for handling missing values and synthetic minority oversampling technique for balancing class distribution across performance levels. Then, to reduce dimensionality and enhance learning efficiency, ensemble-based feature selection approach is devised using symmetric uncertainty, mutual information, F-Score, Gini Index, and Chi-Square. The chosen features are utilized by the proposed attention based cascaded shuffle long term dependent network (ACLNet) model for analyzing the performance of the IT professionals. The proposed model is optimized using the chaotic football team training (ChF2T) algorithm for enhancing the convergence and generalization by integrating chaotic dynamics into the parameter search process. The proposed ChF2T-ACLNet method acquired the accuracy, precision, recall, F-score, Kappa, and computation time of 98.9%, 99.1%, 98.7%, 98.4%, 0.985, and 175 s respectively.