Visualize base learner traces

This function shows how the base learners evolves over the fitting process. The default is to show how the frequency of a single base learner included into the model evolves. Additionally, with the value argument, vectors (e.g. the risk) can be used to show how the base learner specific risk reduction evolves during the fitting process.

Usage

plotBaselearnerTraces(cboost, value = 1, n_legend = 5L)

Arguments

cboost

(Compboost)
A trained Compboost object.

value

(numeric(1L) | numeric(length(cboost$getSelectedBaselearner())))
Value used to show the base learner development w.r.t. to the value.

n_legend

(integer(1L))
Number of colored base learners added to the legend.

Value

ggplot object containing the graphic.

Examples

cboost = Compboost$new(data = iris, target = "Petal.Length",
 loss = LossQuadratic$new())
cboost$addComponents("Sepal.Width")
cboost$addBaselearner("Species", "ridge", BaselearnerCategoricalRidge)
cboost$train(500L)
#>   1/500   risk = 1.4  
#>  12/500   risk = 0.52  
#>  24/500   risk = 0.21  
#>  36/500   risk = 0.13  
#>  48/500   risk = 0.1  
#>  60/500   risk = 0.093  
#>  72/500   risk = 0.088  
#>  84/500   risk = 0.085  
#>  96/500   risk = 0.083  
#> 108/500   risk = 0.081  
#> 120/500   risk = 0.079  
#> 132/500   risk = 0.078  
#> 144/500   risk = 0.077  
#> 156/500   risk = 0.076  
#> 168/500   risk = 0.075  
#> 180/500   risk = 0.074  
#> 192/500   risk = 0.074  
#> 204/500   risk = 0.073  
#> 216/500   risk = 0.073  
#> 228/500   risk = 0.072  
#> 240/500   risk = 0.072  
#> 252/500   risk = 0.072  
#> 264/500   risk = 0.071  
#> 276/500   risk = 0.071  
#> 288/500   risk = 0.071  
#> 300/500   risk = 0.071  
#> 312/500   risk = 0.071  
#> 324/500   risk = 0.07  
#> 336/500   risk = 0.07  
#> 348/500   risk = 0.07  
#> 360/500   risk = 0.07  
#> 372/500   risk = 0.07  
#> 384/500   risk = 0.07  
#> 396/500   risk = 0.069  
#> 408/500   risk = 0.069  
#> 420/500   risk = 0.069  
#> 432/500   risk = 0.069  
#> 444/500   risk = 0.069  
#> 456/500   risk = 0.069  
#> 468/500   risk = 0.069  
#> 480/500   risk = 0.069  
#> 492/500   risk = 0.069  
#> 
#> 
#> Train 500 iterations in 0 Seconds.
#> Final risk based on the train set: 0.069
#> 
plotBaselearnerTraces(cboost)