Alternating Recommendation of Candidate Points
This "multi ensemble" (see details) is comparable to the "single ensemble"
In contrast to round search, models are not alternated from one sequential SPOT step to another.
Instead, all models are used to alternatingly to suggest multiple points for each step. This of course means that SPOT
needs to be configured to use several new candidates in each sequential step (i.e.
spotConfig$seq.design.new.size should be larger than one,
ideally a multiple of the number of models in the ensemble). To determine the order of models, the feedback (e.g. error) of the model is used.
Relevant configuration parameters of this ensemble are:
The function to calculate feedback (e.g. the error) of the models, default is
The number of designs cut off for leave-one-out validation should always be zero for this ensemble, i.e
spotConfig$seq.ensemble.cut.num <- 0
new design points which should be predicted
global list of all options, needed to provide data for calling functions
should always be NULL. Evaluation of existing fits is not implemented for this ensemble.
This is a "multi ensemble", meaning that in every sequential step all models in the ensemble are trained and evaluated.
The target is to actively combine all models responses, to get the best estimate on which candidate points are optimal.
The models used are specified in the
spotConfig list, for instance:
spotConfig$seq.ensemble.predictors = c(spotPredictRandomForest, spotPredictEarth, spotPredictForrester, spotPredictDace)
To specify the settings of each individual model, set:
seq.ensemble.settings = list(list(setting=1),list(setting=2),list(setting=3),list(setting=4))
Any parameters set in each of the corresponding lists (here: 4 individual lists) will overwrite settings in the main
when the concerned model function is called.
returns the list
- M. Friese, M. Zaefferer, T. Bartz-Beielstein, O. Flasch, P. Koch, W. Konen, and B. Naujoks. Ensemble based optimization and tuning algorithms. In F. Hoffmann and E. Huellermeier, editors, Proceedings 21. Workshop Computational Intelligence, p. 119-134. Universitaetsverlag Karlsruhe, 2011.
Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.