predict.bess: make predictions from a "bess" object.

Description

Returns predictions from a fitted "bess" object.

Usage

# S3 method for bess
predict(object, newx, type = c("link", "response"), ...)

Arguments

object

Output from the bess function.

newx

New data used for prediction. If omitted, the fitted linear predictors are used.

type

type = "link" gives the linear predictors for "binomial", "poisson" or "cox" models; for "gaussian" models it gives the fitted values. type = "response" gives the fitted probabilities for "binomial", fitted mean for "poisson" and the fitted relative-risk for "cox"; for "gaussian", type = "response" is equivalent to type = "link"

…

Additional arguments affecting the predictions produced.

Value

The object returned depends on the types of family.

References

Wen, C., Zhang, A., Quan, S. and Wang, X. (2020). BeSS: An R Package for Best Subset Selection in Linear, Logistic and Cox Proportional Hazards Models, Journal of Statistical Software, Vol. 94(4). doi:10.18637/jss.v094.i04.

Examples

Run this code

# NOT RUN {
#-------------------linear model----------------------#
# Generate simulated data
n <- 200
p <- 20
k <- 5
rho <- 0.4
seed <- 10
Tbeta <- rep(0, p)
Tbeta[1:k*floor(p/k):floor(p/k)] <- rep(1, k)
Data <- gen.data(n, p, k, rho, family = "gaussian", beta = Tbeta, seed = seed)
x <- Data$x[1:140, ]
y <- Data$y[1:140]
x_new <- Data$x[141:200, ]
y_new <- Data$y[141:200]
lm.bss <- bess(x, y, method = "sequential")
lambda.list <- exp(seq(log(5), log(0.1), length.out = 10))
lm.bsrr <- bess(x, y, type = "bsrr", method = "pgsection")

pred.bss <- predict(lm.bss, newx = x_new)
pred.bsrr <- predict(lm.bsrr, newx = x_new)

#-------------------logistic model----------------------#
#Generate simulated data
Data <- gen.data(n, p, k, rho, family = "binomial", beta = Tbeta, seed = seed)

x <- Data$x[1:140, ]
y <- Data$y[1:140]
x_new <- Data$x[141:200, ]
y_new <- Data$y[141:200]
logi.bss <- bess(x, y, family = "binomial", method = "sequential", tune = "cv")
lambda.list <- exp(seq(log(5), log(0.1), length.out = 10))
logi.bsrr <- bess(x, y, type = "bsrr", tune="cv",
                 family = "binomial", lambda.list = lambda.list, method = "sequential")

pred.bss <- predict(logi.bss, newx = x_new)
pred.bsrr <- predict(logi.bsrr, newx = x_new)

#-------------------coxph model----------------------#
#Generate simulated data
Data <- gen.data(n, p, k, rho, family = "cox", beta = Tbeta, scal = 10)

x <- Data$x[1:140, ]
y <- Data$y[1:140, ]
x_new <- Data$x[141:200, ]
y_new <- Data$y[141:200, ]
cox.bss <- bess(x, y, family = "cox", method = "sequential")
lambda.list <- exp(seq(log(5), log(0.1), length.out = 10))
cox.bsrr <- bess(x, y, type = "bsrr", family = "cox", lambda.list = lambda.list)

pred.bss <- predict(cox.bss, newx = x_new)
pred.bsrr <- predict(cox.bsrr, newx = x_new)

#-------------------group selection----------------------#
beta <- rep(c(rep(1,2),rep(0,3)), 4)
Data <- gen.data(200, 20, 5, rho=0.4, beta = beta, seed =10)
x <- Data$x
y <- Data$y

group.index <- c(rep(1, 2), rep(2, 3), rep(3, 2), rep(4, 3),
                rep(5, 2), rep(6, 3), rep(7, 2), rep(8, 3))
lm.group <- bess(x, y, s.min=1, s.max = 8, type = "bss", group.index = group.index)
lm.groupbsrr <- bess(x, y, type = "bsrr", s.min = 1, s.max = 8, group.index = group.index)

pred.group <- predict(lm.group, newx = x_new)
pred.groupbsrr <- predict(lm.groupbsrr, newx = x_new)

# }