Adaptive-CSSA: adaptive-chicken squirrel search algorithm driven deep belief network for student stress-level and drop out prediction with MapReduce framework

Stress is correlated with various illnesses that include diabetes, depression, and other chronic diseases and plays an important role in the emotional and physical well-being. In past years, number of student dropout from educational institute is rapidly maximizing. However, high rate of student dropout has been a major issue in various institutions. Dropout and stress prediction has received more attention in recent years. Previous literature studies applied machine learning algorithms to recognize the dropout and stress level of students, but there exist an issue of low accuracy and it leads to misidentification at learners. To overcome such problems and to create more accurate prediction result, a proposed method named Adaptive Chicken Squirrel Search Algorithm on the basis of Deep Belief Network (Adaptive-CSSA based DBN) is developed to predict the dropout and stress level of students based on the student performance behavior. The proposed prediction mechanism is designed with MapReduce framework by considering the mapper and the reducer functions. Here, the feature selection strategy is employed to select the unique features from student data that contains detailed information of students. Moreover, the proposed technique attained minimum MSE, RMSE, and MAPE as 0.025, 0.157, and 0.126 for dropout prediction and obtained lower MSE, RMSE, and MAPE value of 0.150, 0.387, and 0.326 for stress level prediction with student performance dataset.