Student academic performance prediction with MapReduce using the optimised deep belief network

The emergence of learners count in the e-learning infrastructures has inspired researchers to carry out data-driven learning assessments for learning performance enhancement among students. A new technique, namely chicken squirrel search algorithm-based deep belief network (CSSA-DBN) is devised with MapReduce for predicting the student academic performance. Here, the details of students are attained, which performs pre-processing with log transformation to make it suitable for improved processing. Moreover, the mappers perform the selection of features with a correlation-based Tversky index wherein the Pearson correlation coefficient and Tversky index are integrated to choose imperative features. In addition, the reducers perform student performance prediction with a deep belief network (DBN), trained using the proposed chicken squirrel search algorithm (CSSA). The proposed CSSA is devised by blending chicken swarm optimisation (CSO) and squirrel search algorithm (SSA).