gbm.subset {gbm.auto}

Subset gbm.auto input datasets to 2 groups using the partial deviance plots

Description

Set your working directory to the output folder of a gbm.auto/gbm.loop run. This function returns the variable value corresponding to the 0 value on the lineplots, which should be the optimal place to split the dataset into 2 subsets, low and high, IF the relationship doesn't cross 0 more than once. Function is similarly useful to quickly get the 0-point value in these cases, i.e. where values below are detrimental, values above beneficial (check plots though)

Usage

gbm.subset(x, fams = c("Bin", "Gaus"), loop = FALSE)

Arguments

x

Vector of variable names.

fams

Vector of statistical data distribution family names to be modelled by gbm.

loop

Is the folder a gbm.loop output?

Details

loop varnames are BinLineLoop_VAR.csv & GausLineLoop_VAR.csv normal varnames are Bin_Best_line_VAR.csv & Gaus_Best_line_VAR.csv

Just use average between the last negative & first positive point unless any points fall on zero

Value

a list of breakpoint values which datasets can be subsetted using.

Author(s)

Simon Dedman, [e-mail]

Examples

# Not run: requires completed gbm.auto run.
# having run gbm.auto (with linesfiles=TRUE), set working directory there
data(samples)
gbm.subset(x = names(samples[c(4:8, 10)]), fams = c("Bin", "Gaus"))