Statistical Learning for Duration Analysis, with application to missing persons data in Montana.
Cox PH output: Reference level of categorical covariates denoted reference and have estimated HR of 1Abstract
This study explored the use of Random Survival Forests (RSF) to predict the time to resolution of missing persons cases in Montana. RSF was compared to traditional methods like the Cox proportional hazards model and Kaplan-Meier curves. While RSF showed promise in terms of prediction error, its performance was comparable to the Cox model, especially over time. Important factors influencing resolution time, as identified by RSF, included age, region, reservation status, and race.
Type
