R Language Survival analysis Random Forest Survival Analysis with randomForestSRC
Example
Just as the random forest algorithm may be applied to regression and classification tasks, it can also be extended to survival analysis.
In the example below a survival model is fit and used for prediction, scoring, and performance analysis using the package randomForestSRC from CRAN.
require(randomForestSRC)
set.seed(130948) #Other seeds give similar comparative results
x1 <- runif(1000)
y <- rnorm(1000, mean = x1, sd = .3)
data <- data.frame(x1 = x1, y = y)
head(data)
x1 y 1 0.9604353 1.3549648 2 0.3771234 0.2961592 3 0.7844242 0.6942191 4 0.9860443 1.5348900 5 0.1942237 0.4629535 6 0.7442532 -0.0672639
(modRFSRC <- rfsrc(y ~ x1, data = data, ntree=500, nodesize = 5))
Sample size: 1000 Number of trees: 500 Minimum terminal node size: 5 Average no. of terminal nodes: 208.258 No. of variables tried at each split: 1 Total no. of variables: 1 Analysis: RF-R Family: regr Splitting rule: mse % variance explained: 32.08 Error rate: 0.11
x1new <- runif(10000)
ynew <- rnorm(10000, mean = x1new, sd = .3)
newdata <- data.frame(x1 = x1new, y = ynew)
survival.results <- predict(modRFSRC, newdata = newdata)
survival.results
Sample size of test (predict) data: 10000 Number of grow trees: 500 Average no. of grow terminal nodes: 208.258 Total no. of grow variables: 1 Analysis: RF-R Family: regr % variance explained: 34.97 Test set error rate: 0.11
