Performance tuning options for rxDForest/rxDTree
|
Goal |
Options |
|
Performance |
xVal              = 0     # (rxDTree) number of cross-validations during fittingcomputeOobError   = -1     # (rxDForest) do not compute OOB prediction errorsscheduleOnce      = TRUE  # launch via rxExec (if modest data; best for Hadoop)                          # Requires nTree = #trees / #tasksblocksPerRead     = large  # if the input is an XDF file – Warning: reduce this if memory consumption is an issueverbose          = 0     # minimize console outputreportProgress    = 0     # dittouseSparseCube     = FALSE  # sparse cube is a bit slower so avoid if possible |
