Payesh (Oct 2024)
The application of survival forests in the study of the most important determinants of the first marriage survival of divorced women with children
Abstract
Objective(s): Divorce as an important social harm has always been the focus of investigators and policymakers as it affects women compared to males. Therefore, considering the importance and prominent role of women in the family and society, the present article aimed to examine the most important determinants of the first marriage survival of divorced women with children. Methods: In a cross-sectional survey conducted by the Civil Registry Organization in 2017-2018, the information of those who referred to the offices of the provincial centers for divorce registration was collected using a questionnaire. Considering the large number of predictors and the ineffectiveness of classic survival analysis methods in big data modeling, the present study investigated the most important factors, including women and their spouses, their families and provincial macro variables, affecting the first marriage survival of 756 women with children using survival forests using R software. Results: Based on the highest value of Harrell's coordination index (0.8412), the lowest mean prediction error (0.0885) and the lowest value of integrated Brier score (0.038), the algorithm of random survival forest with log rank split rule (RSF1) in investigating factors affecting the first marriage survival of these women was more efficient. The findings showed that based on variable importance and minimum depth indicators, the first child’s age was the most important variable in examining their first marriage survival. Conclusions: Since big data are analyzed in many medical and social studies, survival forests can be used as an efficient method to identify the most important predictors and reduce their dimensions, and then use classical survival analysis methods for modeling.
