Suicide prospective prediction based on pattern analysis of suicide factor

• Main Authors: Dawood, Aya Qusay, Mostafa, Salama A., Mahdin, Hairulnizam, Pramudya, Gede, Kasim, Shahreen, Alkhayyat, Ahmed, Ismail, Saidatul Akmar, Arshad, Mohammad Syafwan
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Algorithms; Decision trees; Machine learning; Mental disorders; Neural networks; Pattern analysis; Suicides & suicide attempts
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0188323

People who seem unable to make it through life may make the tragic and saddening decision to end their lives. The nation’s backbone is its youth; with their health and bravery, they have the power to influence the future of society. Nevertheless, youth have a higher percentage of attempting suicide and it is due to different reasons including physical disorders, mental disorders and substance use disorders. Support from family, friends, and society is essential in preventing individuals from making such tragic mistakes. Different studies attempt to acquire suicide indicators and measure risk factors to take preventive actions against potential suicidal situations. This paper presents the comparative analysis of four machine learning algorithms: Random Forest (RF), Neural Network (NN), Logistic Regression (LR), and Decision Tree (DT) in predicting suicide risk rate. The result of this study shows that the RF has outperformed the other three algorithms with an average accuracy of 92%. The DT and NN show an average accuracy of 88% and 80%, respectively, while the LR shows the lowest accuracy of 69%.