prim.S3methods: S3 methods for PRIM for multivariate data

In prim: Patient Rule Induction Method (PRIM)

S3 methods for PRIM for multivariate data

Description

S3 methods PRIM for multivariate data.

Usage

## S3 method for class 'prim'
predict(object, newdata, y.fun.flag=FALSE, ...)
## S3 method for class 'prim'
summary(object, ..., print.box=FALSE)

Arguments

object	object of class prim
newdata	data matrix
y.fun.flag	flag to return y value of PRIM box rather than box label. Default is FALSE.
print.box	flag to print out limits of all PRIM boxes. Default is FALSE.
...	other parameters

Details

–The predict method returns the value of PRIM box number in which newdata are located.

–The summary method displays a table with three columns: box-fun is the y value, box-mass is the mass of the box, threshold.type is the threshold direction indicator: 1 = ">= threshold", -1 = "<=threshold". Each box corresponds to a row. The second last row marked with an asterisk is the box which collates the remaining data points not belonging to a specific PRIM box. The final row is an overall summary, i.e. box-fun is the overall mean of y and box-mass is 1.

Examples

data(quasiflow)
qf <- quasiflow[1:1000,1:2]
qf.label <- quasiflow[1:1000,3]*quasiflow[1:1000,4]

qf.prim <- prim.box(x=qf, y=qf.label, threshold=c(0.3, -0.1), threshold.type=0,
  verbose=TRUE)
## verbose=TRUE prints out extra informaton about peeling and pasting

summary(qf.prim)
predict(qf.prim, newdata=c(0.6,0.2))

## using median insted of mean for the response y

qf.prim2 <- prim.box(x=qf, y=qf.label, threshold=c(0.5, -0.2),
  threshold.type=0, y.fun=median)
summary(qf.prim2)
predict(qf.prim2, newdata=c(0.6,0.2))

prim documentation built on Jan. 7, 2023, 1:24 a.m.