This study set out to present a predictive model for suicidal ideation of the elderly using a random forest and identified the characteristics of predictors. The Survey on the Elderly data, from 2014, 2016, and 2020 conducted by the Ministry of Health and Welfare, was used in the empirical analysis. The subjects of the study were 18,258 elderlies. Predictive models were estimated using 31 factors, which were to predict suicidal thoughts of the elderly reported in previous studies and variables to be added in subsequent studies, and random forests machine learning algorithm. The evaluation of predictive models was shown to be an accuracy of 80.70%, a sensitivity of 26.46%, and a specificity of 96.21%, respectively. The relative importance of the predictors was that the importance of variables in terms of economic conditions was generally higher than the others. This study tries to reflect the risk assessment of suicide attempts and discuss intervention methods for these variables, and it is thought that countermeasures are required to prevent suicidal thoughts or attempts.
1. Use Google Colab
!git clone https://github.com/Skymind24/2022-ML1-project.git
2. (Not Needed. Final dataset is in data folder)
Run data_preparation.ipynb
3.
Run main.ipynb