Opened 10 years ago

Closed 10 years ago

#1450 closed feature request (done)

Support for ensemble solutions of data analysis models

Reported by: gkronber Owned by: gkronber
Priority: highest Milestone: HeuristicLab 3.3.5
Component: Problems.DataAnalysis Version: 3.3.5
Keywords: Cc:

Description

It should be possible to combine multiple models for a data analysis problem into a solution ensemble. For classification different voting schemes should be implemented. For regression different types of estimated values aggregation functions should be implemented.

Cross-validation should produce solution ensembles that combine the models of all folds and the training and test partitions of each model should be taken into consideration when calculating the estimated values for training and test of the ensemble solution.

Change History (21)

comment:1 Changed 10 years ago by gkronber

  • Status changed from new to accepted

comment:2 Changed 10 years ago by gkronber

Created branch for implementation of ticket with r5815.

This feature was partially implemented in ticket #1418 but will not be finished before the 3.3.4 release.

comment:3 Changed 10 years ago by gkronber

r5816: added preliminary implementation for solution ensemble support.

comment:4 Changed 10 years ago by gkronber

  • Status changed from accepted to assigned

comment:5 Changed 10 years ago by gkronber

  • Priority changed from medium to highest

comment:6 Changed 10 years ago by gkronber

  • Status changed from assigned to accepted

comment:7 Changed 10 years ago by gkronber

  • Milestone changed from HeuristicLab 3.3.x Backlog to HeuristicLab 3.3.5
  • Version changed from 3.3.3 to 3.3.4

comment:8 Changed 10 years ago by gkronber

  • Summary changed from Support for ensemble solutions of data analysis models. to Support for ensemble solutions of data analysis models

comment:9 Changed 10 years ago by gkronber

r6184: merged r5816 from the branch and implemented first version of ensemble solutions for regression. The ensembles are only produced by cross validation.

comment:10 Changed 10 years ago by gkronber

r6185: deleted obsolete branch for ensemble solutions.

comment:11 Changed 10 years ago by gkronber

r6238: adapted views for regression solution to work for ensembles of regression solutions as well.

comment:12 Changed 10 years ago by gkronber

r6239: implemented support for ensemble solutions for classification.

comment:13 Changed 10 years ago by gkronber

r6250: added flattening of sub-results in ensemble solutions for classification and regression in the cross-validation.

comment:14 Changed 10 years ago by gkronber

r6252: small change in line chart to remove white space between training and test partitions.

comment:15 Changed 10 years ago by gkronber

r6254: fixed bugs in calculation of estimated values in ensemble solutions for regression and classification.

comment:16 Changed 10 years ago by gkronber

r6255: implemented changes in estimated values views for regression and classification solutions specific to ensemble solutions.

comment:17 Changed 10 years ago by gkronber

r6302: fixed cloning bug and a problem in the regression line chart view.

comment:18 Changed 10 years ago by gkronber

r6520: implemented menu item to merge all data analysis solutions in a run collection into ensemble solutions and fixed a few bugs related to ensemble solutions.

comment:19 Changed 10 years ago by gkronber

  • Owner changed from gkronber to mkommend
  • Status changed from accepted to reviewing

comment:20 Changed 10 years ago by mkommend

  • Owner changed from mkommend to gkronber
  • Status changed from reviewing to readytorelease

Thanks for implementing this. Tested the functionality and it works as expected. Further development will be tracked in other tickets.

comment:21 Changed 10 years ago by swagner

  • Resolution set to done
  • Status changed from readytorelease to closed
  • Version changed from 3.3.4 to 3.3.5
Note: See TracTickets for help on using tickets.