Transform and Re-Transform the Target Variable — mlr_graphs

Wraps a Graph that transforms a target during training and inverts the transformation during prediction. This is done as follows:

Specify a transformation and inversion function using any subclass of PipeOpTargetTrafo, defaults to PipeOpTargetMutate, afterwards apply graph.
At the very end, during prediction the transformation is inverted using PipeOpTargetInvert.
To set a transformation and inversion function for PipeOpTargetMutate see the parameters trafo and inverter of the param_set of the resulting Graph.
Note that the input graph is not explicitly checked to actually return a Prediction during prediction.

pipeline_targettrafo(
  graph,
  trafo_pipeop = PipeOpTargetMutate$new(),
  id_prefix = ""
)

Arguments

graph	`PipeOpLearner` \| `Graph` A `PipeOpLearner` or `Graph` to wrap between a transformation and re-transformation of the target variable.
trafo_pipeop	`PipeOp` A `PipeOp` that is a subclass of `PipeOpTargetTrafo`. Default is `PipeOpTargetMutate`.
id_prefix	`character(1)` Optional id prefix to prepend to `PipeOpTargetInvert` ID. The resulting ID will be `"[id_prefix]targetinvert"`. Default is `""`.

Value

Graph

Examples

library("mlr3")

tt = pipeline_targettrafo(PipeOpLearner$new(LearnerRegrRpart$new()))
tt$param_set$values$targetmutate.trafo = function(x) log(x, base = 2)
tt$param_set$values$targetmutate.inverter = function(x) list(response = 2 ^ x$response)

# gives the same as
g = Graph$new()
g$add_pipeop(PipeOpTargetMutate$new(param_vals = list(
  trafo = function(x) log(x, base = 2),
  inverter = function(x) list(response = 2 ^ x$response))
  )
)
g$add_pipeop(LearnerRegrRpart$new())
g$add_pipeop(PipeOpTargetInvert$new())
g$add_edge(src_id = "targetmutate", dst_id = "targetinvert",
  src_channel = 1, dst_channel = 1)
g$add_edge(src_id = "targetmutate", dst_id = "regr.rpart",
  src_channel = 2, dst_channel = 1)
g$add_edge(src_id = "regr.rpart", dst_id = "targetinvert",
  src_channel = 1, dst_channel = 2)