martPred {mseq}R Documentation

predict using MART

Description

Get the predicted log sequencing preferences by the trained MART model, and adjust them so that read TT....T have value 0.

Usage

martPred(train.gbm, newdata, n.trees = 2000)

Arguments

train.gbm the trained gbm object
newdata the data that we want to predict
n.trees number of trees to be considered

Value

a numeric vector of predicted log preferences

Examples

 # read and expand the data
 data(g1_part) # for real data, please use read.csv, like g1 <- read.csv("g1.csv")
 data <- expData(g1_part, 2, 3) # here the surrounding sequences is only of length 5. In real datasets, it should be larger.
 
 # train and predict by MART
 train.data <- data[data$index < 6, ]
 test.data <- data[data$index >= 6, ]
 train.mart <- martTrain(train.data, interaction.depth = 2, n.trees = 100)
 pred.pref <- exp(martPred(train.mart, test.data, n.trees = 100))

[Package mseq version 1.1 Index]