Skip to contents

PDA projection pursuit strategy.

pp_pda.Rd

Creates a Penalized Discriminant Analysis (PDA) projection pursuit strategy for use with pptr or pprf.

Usage

pp_pda(lambda = 0)

Arguments

lambda

A regularization parameter between 0 and 1. If lambda = 0, the model uses Linear Discriminant Analysis (LDA). If lambda > 0, the model uses Penalized Discriminant Analysis (PDA).

Value

A pp_strategy object.

See also

pptr, pprf, vars_uniform, vars_all, cutpoint_mean_of_means

Examples

# PDA with lambda = 0.5
pp_pda(0.5)
#> $name
#> [1] "pda"
#> 
#> $display_name
#> [1] "PDA"
#> 
#> $lambda
#> [1] 0.5
#> 
#> attr(,"class")
#> [1] "pp_strategy"

# Use with pptr
pptr(Species ~ ., data = iris, pp = pp_pda(0.5))
#> 
#> Call: pptr(formula = Species ~ ., data = iris, pp = pp_pda(0.5))
#> 
#> Projection-Pursuit Oblique Decision Tree:
#> If ([ 0 -0.04 0.03 0.03 ] * x) < 0.01580044:
#>   Predict: setosa 
#> Else:
#>  If ([ 0 0.03 -0.06 -0.15 ] * x) < -0.4503323:
#>    Predict: virginica 
#>  Else:
#>    Predict: versicolor 
#> 

# Use with pprf
pprf(Species ~ ., data = iris, pp = pp_pda(0.5), vars = vars_uniform(n_vars = 2))
#> 
#> Random Forest of Projection-Pursuit Oblique Decision Trees
#>   Call:        pprf(formula = Species ~ ., data = iris, pp = pp_pda(0.5), vars = vars_uniform(n_vars = 2))
#>   Trees:       100
#>   Mode:        classification
#>   Group names: setosa, versicolor, virginica
#>   Formula:     Species ~ Sepal.Length + Sepal.Width + Petal.Length + Petal.Width -     1
#>