HFIPO-DPNN: A Framework for Predicting the Dropout of Physically Impaired Student from Education

Abstract

Education plays a significant role in individuals’ development and the economic growth of developing countries like India. Dropout of students from their studies is the major concern for any order of education. Some models for predicting the dropout of students are developed with several factors. Many of them lacked consistency as they backed their studies with the academic performance of the students. Especially, for those students who suffered from physical impairment, the dropout depends on several external factors. Hence, this work proposes a novel HFIPO-DPNN to predict the student dropout rooted in the previous semester’s marks. The proposed model enclosed the hybrid firefly and improved particle swarm algorithm to optimize the feature selection that influences the dropout of hearing-impaired students. The optimized feature data are used to predict the dropout with the novel DPNN. The optimized data was split and used for training the DPNN. The testing data is used to evaluate the performance of the proposed framework. The attributes used for predicting the student dropout are Family Size, Subject, Medium of Instruction, and so on. The data must be collected from 250 physically impaired children belonging to ITI institute, Bangalore. The outcome of the proposed framework is evaluated on several metrics. The accuracy of the proposed model is about 99.02%. The HFIPO-DPNN framework can be enhanced for predicting the dropout for students with other disabilities. The optimization showed that factors influencing education other than familial factors are to be considered in the prediction of dropout.