Predictive Analysis of Mental Health Using Machine Learning for Depression Prediction

This work aims to predict depression based on diverse data using Machine Learning Algorithms. The designed model seeks to identify early indicators of depression, providing a potential tool for proactive intervention and support in mental health by analyzing patterns in behavioral, physiological, and contextual data. Machine learning algorithms, namely decision trees, extra trees, XGBoost, Stochastic gradient descent, grid search CV, Stacking, and Voting classifiers, etc., are used to predict depression in the early stage. This study emphasizes integrating machine learning techniques to enhance predictive accuracy and contribute to developing accessible and timely depression detection systems. The F1 score was added, which helped to identify the best machine learning algorithm among the ones applied. We have achieved an accuracy of 92 % with random forest, which is 3% higher than the work previously done in RF. We also achieved a 0.99 F1 score using Linear SVM.