DefinePK

Abstract

Mental health illness is very harmful for the individual and the surrounding people.It changes persons opinion and finds difficulty in positive thought process. The main aim is to implement a hybrid approach for prediction of mental health. This research includes the analysis of two different models One is naive byes and another is CNN . Here ,we have done feature engineering and model building with the help of both models to follow a combined approach . Datasets used for training and testing are : Student_Behaviour (records 40960 ,attributes 25), Student mental health (records 101, attributes 11) ) which  applied to the task of prediction of mental health illness. They were tested for classification performance measures like accuracy ,precision ,F1- score, Recall. A hybrid model which includes feature extraction using Naive Bayes and model building using CNN has given highest accuracy (80.85% ) for the dataset Student_Behaviour. So ,By using this method we make predictions on our real world unlabelled student data which is collected manually . Experimental results show 9.8% students are mentally unsafe and the remaining 90.2% are mentally safe among collected data.