Random forests with survival data

We often want to model the time it takes for some event to occur (e.g. death, failure, etc). In medical research, we typically do this using the Cox proportional hazards model, or some flavor of parametric survival model. Linear models like these are especially useful because model parameters can often be meaningfully interpreted. However, when we are primarily concerned with prediction of future events, the performance of linear models is often outclassed by other algorithms, such as Random Forests, an ensemble method for regression/classification trees. In this meet-up we will demonstrate the use of random survival forests using the R package randomForestSRC. https://cran.r-project.org/web/packages/randomForestSRC/randomForestSRC.pdf