Comparative Analysis of Machine Learning Algorithms for Mental health issue classification

Abstract

In today's fast paced life, stress is a very common problem but if this problem persists then it takes the form of a mental illness. The rising incidence of mental health conditions and the expanding amount of digital behavioral data have drawn a lot of attention to machine learning (ML)-based mental illness detection.  In order to classify and predict mental health conditions like depression, anxiety, stress, personality disorder, bipolar disorder and suicidal this study compares a number of popular machine learning algorithms, including Random Forests, Naïve Bayes (NB), Multinomial NB, Decision Tree and Logistic Regression. Accuracy, precision, recall and F1-score are the metrics used to assess each algorithm using a structured dataset. The findings show that ensemble techniques like decision tree achieve highest accuracy which is 78%.