Predicting Student Performance and its Outcomes in the Modern Education Systems Using Attention Based BiRNN and CNN Method

Education is essential for the development and prosperity of every nation. The minds and hearts of those who receive an education are molded in the process. It empowers people and makes them want to achieve their goals. Traditional higher education functions similarly to the brick-and-mortar business model in that students acquire a uniform education by physically attending classrooms on a regular (full- or part-time) basis. However, the standard educational system has many problems, and major improvements are on the horizon. One possible aspect of the future educational system is the proliferation of online education. The proposed strategy makes use of feature selection, preprocessing, and model training. Field extraction, data cleaning and filtering, user identification, and data synthesis are among preprocessing techniques employed by our team. By locating the correlation matrix's eigenvalues and eigenvectors, PCA can be used for feature selection. After deciding on a set of features, the models are trained via BiRNN-CNN. CNN and BiRNN are two common alternatives, although the proposed method surpasses them both. The proposed method was successful to the tune of 98.23%.