visualpred (version 0.1.0)

famdcontourlabel: Outliers in Contour plots and FAMD function for classification modeling

Description

This function adds outlier marks to famdcontour using ggrepel package.

Usage

famdcontourlabel(
  dataf = dataf,
  Idt = "",
  inf = 0.1,
  sup = 0.9,
  cutprob = 0.5,
  ...
)

Arguments

dataf

data frame.

Idt

Identification variable, default "", row number

inf, sup

Quantiles for x,y outliers

cutprob

cut point for outliers based on prob.estimation error

...

options to be passed from famdcontour

Value

A list with the following objects:

graph1_graph6

plots for dimension outliers

graph7_graph12

plots for fit outliers

Details

An identification variable can be set in Idt parameter. By default, number of row is used. There are two source of outliers: i) outliers in the two FAMD dimension space, where the cutpoints are set as quantiles given (inf=0.1 and sup=0.9 in both dimensions by default) and ii) outliers with respect to the fitted probability. The dependent variable is set to 1 for the mimority class, and 0 for the majority class. Points considered outliers are those for which abs(vardep-fittedprob) excede parameter cutprob.

Examples

# NOT RUN {
data(breastwisconsin1)
dataf<-breastwisconsin1
listconti=c( "clump_thickness","uniformity_of_cell_shape","mitosis")
listclass=c("")
vardep="classes"
result<-famdcontourlabel(dataf=dataf,listconti=listconti,
listclass=listclass,vardep=vardep)
# }