YohannLeFaou/sword: Random forest with IPC weights for survival analysis

sword is a package which implements the methology developed in (citer article). The main function of sword is sw_reg and allows to fit random forest and gam model to model a right censored variable, using Inverse Probability of Censoring Weighting (IPCW). Different types of weights are available (Kaplan-Meier, Cox, random surv forest) and custom weights may be provided. cox_reg and rsf_reg are functions which implement the 2 benchmarks based on Cox model and RSF model studied in (citer article). These 3 sword functions are wrappers to other R packages which implement classical regression algorithms.

install.packages("remotes")
remotes::install_github("YohannLeFaou/sword")