Xsurv.cv: This function can use cv to automatically tuning the paramter...

In topycyao/Xsurv: A package used to model survival data from advanced machine learning algorithms

This function can use cv to automatically tuning the paramter to fit the survival model for xgboost and lightgbm.

Description

This function can use cv to automatically tuning the paramter to fit the survival model for xgboost and lightgbm.

Usage

Xsurv.cv(
  datax,
  datay,
  top_n = NULL,
  option = c("defaut", "xgb", "lgb", "gbm", "rf"),
  method = c("defaut", "pl", "C"),
  search = c("rd", "grid"),
  nfolds = 5,
  cvfrac = 0.8,
  rdtime = 10,
  nround = NULL,
  Lambda = NULL,
  Alpha = NULL,
  Eta = NULL,
  early_stopping_rounds = NULL,
  cp = NULL,
  maxdpth = NULL
)

Arguments

datax	X data set
datay	Y data set including time and event status
top_n	number of top features for survival tree fitting
option	model fitting option,defaut is xgb
method	fitting metohd,defaut is 'pl' means using loss function:coxph likelihood
nfolds	number of folds for crossvalidation
early_stopping_rounds	force a stopping round
number	of rounds
lambda	l1 penalty parameter
alpha	l2 penalty parameter
eta	learning rate

Value

a list object containing:model,cindex,tree,SHAP and risk

Examples

#example
library(survival)
library(Xsurv)
data(lung)

mydata<-(lung[,-1])
mydata[,2]<-mydata[,2]-1
length(mydata[,1])
names(mydata)<-colnames(mydata)
datay_train<-mydata[1:180,c(1,2)]
datax_train<-mydata[1:180,-c(1,2)]
datay_test<-mydata[181:228,c(1,2)]
datax_test<-mydata[181:228,-c(1,2)]
xs<-Xsurv(datax_train,datay_train,top_n = 5,cp=0.01)

#xs<-Xsurv.cv(datax_train,datay_train,top_n=5)
xm<-xs$model
xtree<-xs$tree
x_ctree<-xtree$tree2
#plot(x_ctree)
shap=xs$SHAP

shap

risk=xs$risk
fit=risk$fit

#plot(fit)
#prediction

pre_time<-pre<-Xsurv_predict(xm,datax_train,datay_train,datax_test)

#predict survival probabilty

pre_x<-Xsurv_predict_sv(xm,datax_train,datay_train,datax_test[1,])

plot(pre_x)

topycyao/Xsurv documentation built on Aug. 6, 2022, 9:06 p.m.