Suicidal Risk Prediction Using Catboost Classifier: A Comparative Study of Six Machine Learning Classifiers

This study employs machine learning classifiers, including Logistic Regression, Random Forest Classifier, Extra Trees Classifier, XGB Classifier, LGBM Classifier, and CatBoost Classifier, to predict suicidal mental health. By employing a thorough methodology, this research endeavours to ascertain the classifier that is most efficient in performing this crucial duty. CatBoost Classifier demonstrated superior performance among the classifiers that were assessed, attaining a training accuracy of 84% and a testing accuracy of 73%. The findings highlight the effectiveness of the classifier in reliably differentiating between cases of depression and Suicide, showcasing its potential usefulness in initiatives aimed at preventing suicide. Although there was a marginal decline in test accuracy when compared to the accuracy achieved during training, the CatBoost Classifier still exhibited robust predictive capabilities, underscoring its dependability in practical situations. The results of this research make a valuable contribution to the field of mental health predictive modelling and underscore the criticality of employing machine learning methods in the realm of suicide prevention. Additional investigation is necessary in order to verify the CatBoost Classifier's scalability and efficacy across a wide range of populations and contexts.