Early Prediction on Depression Based On Text Classification Method of Machine Learning

Abstract

Mental illness includes emotional, psychological, and social well-being. Mental illness starts from a mild depression and it could lead to other serious types of mental illness. In 2017, a study estimates that 792 million people lived with mental illness which takes about 10.7% globally. Tools such as questionnaire is normally used by the clinician in diagnosing the early symptom of depression and there are various type of questionnaire that is used to predict various level of depression. In this study, the Diagnostic and Statistical Manual of Mental Disorders, version 5 (DSM-5) is used with several modification and were sent to 113 participants through the platform of media social. Findings show that, females suffered most from depression with 65% rather than males with only 35%. Besides, based on the age factors, adults that age between 36 to 45 years old suffered from mental illness more than other group of age with the percentage of 25%. Then, experiment also have been executed with three Machine Learning methods namely, Gauss Naïve Bayes, Random Forest and Decision Tree to observe the accuracy of these methods in doing the classification in predicting the depression into depress and not depress. The finding shows that Decision Tree shows the highest accuracy with 86.96% when compared to Gauss Naives Bayes and Random Forest with the accuracy of 82.6% and 78.26% respectively.