Pre-Assessment Question Paper Analysis and Early Result Predicting Using EDM

Under competitive examination it is of paramount importance that students can be provided with customized question papers and that student performance can be predicted reliably. As for examinations, it is now the most important task to formulate new examination questions and predict the success of the students. In this project, machine learning models, in particular, LightGBM, are used for analysis of the historical data on student performance with question banks to generate the targeted question papers. In this system, both behavioral features and content related features are utilized to estimate the probability that a student could get the right answer with an improved AUC score of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 8 0}$</tex>. The model has been implemented in a real-world prediction pipeline system in which adaptive testing is possible and individualized learning can be achieved. The results show considerable advancement in exam preparation efficiency, comprehension of student achievement, and the automatic generation of individualized question papers, which enhances