Skip to contents

Create a new Graph for a classification Task to perform "One vs. Rest" classification.

All input arguments are cloned and have no references in common with the returned Graph.

Usage

pipeline_ovr(graph)

Arguments

graph

Graph
Graph being wrapped between PipeOpOVRSplit and PipeOpOVRUnite. The Graph should return NULL during training and a classification Prediction during prediction.

Value

Graph

Examples

library("mlr3")

task = tsk("wine")

learner = lrn("classif.rpart")
learner$predict_type = "prob"

# Simple OVR
g1 = pipeline_ovr(learner)
g1$train(task)
#> $ovrunite.output
#> NULL
#> 
g1$predict(task)
#> $ovrunite.output
#> <PredictionClassif> for 178 observations:
#>     row_ids truth response     prob.1     prob.2     prob.3
#>           1     1        1 0.93301153 0.04437722 0.02261125
#>           2     1        1 0.93301153 0.04437722 0.02261125
#>           3     1        1 0.93301153 0.04437722 0.02261125
#> ---                                                        
#>         176     3        3 0.08896784 0.04157376 0.86945841
#>         177     3        3 0.08896784 0.04157376 0.86945841
#>         178     3        3 0.01729144 0.04484461 0.93786395
#> 

# Bagged Learners
gr = po("replicate", reps = 3) %>>%
  po("subsample") %>>%
  learner %>>%
  po("classifavg", collect_multiplicity = TRUE)
g2 = pipeline_ovr(gr)
g2$train(task)
#> $ovrunite.output
#> NULL
#> 
g2$predict(task)
#> $ovrunite.output
#> <PredictionClassif> for 178 observations:
#>     row_ids truth response     prob.1     prob.2     prob.3
#>           1     1        1 0.93801099 0.05417103 0.00781798
#>           2     1        1 0.93801099 0.05417103 0.00781798
#>           3     1        1 0.93801099 0.05417103 0.00781798
#> ---                                                        
#>         176     3        3 0.21422882 0.04551151 0.74025967
#>         177     3        3 0.21422882 0.04551151 0.74025967
#>         178     3        3 0.03695298 0.05577925 0.90726778
#> 

# Bagging outside OVR
g3 = po("replicate", reps = 3) %>>%
  pipeline_ovr(po("subsample") %>>% learner) %>>%
  po("classifavg", collect_multiplicity = TRUE)
g3$train(task)
#> $classifavg.output
#> NULL
#> 
g3$predict(task)
#> $classifavg.output
#> <PredictionClassif> for 178 observations:
#>     row_ids truth response     prob.1     prob.2      prob.3
#>           1     1        1 0.94066508 0.05111255 0.008222368
#>           2     1        1 0.94066508 0.05111255 0.008222368
#>           3     1        1 0.94066508 0.05111255 0.008222368
#> ---                                                         
#>         176     3        3 0.11028161 0.04653909 0.843179301
#>         177     3        3 0.11028161 0.04653909 0.843179301
#>         178     3        3 0.03724958 0.05160886 0.911141558
#>