Free cookie consent management tool by TermsFeed Policy Generator

Timeline


and

09/28/09:

17:00 Changeset [2398] by mkommend

adapted behaviour of CEDMA server to the new modeling database (ticket #759)

16:17 Changeset [2397] by mkommend

corrected ModelingResultCalculators for TimeSeriesPrognosis (ticket #755)

14:36 Ticket #768 (CEDMA engines should use a limited number of trainingsamples to ...) closed by gkronber
fixed: Fixed with r2396.
14:36 Changeset [2396] by gkronber

Fixed #768 (CEDMA engines should use a limited number of trainingsamples to generate more models in less time)

14:32 Ticket #768 (CEDMA engines should use a limited number of trainingsamples to ...) created by gkronber
14:26 Changeset [2395] by gkronber

Increased the limit for termination of GP engines when overfitting is likely. #762 (ValidationQuality improvement progress as additional stopping criterion for default StructId Engines)

13:32 Ticket #767 (Save functionality for CEDMA settings) created by mkommend
The adjuste CEDMA settings (targets, allowed inputs, algorithms, …
10:56 Changeset [2394] by mkommend

removed unnecessary class !IProblem from modeling database (ticket #759)

09/27/09:

17:28 Ticket #765 (GP Predictor should output NaN if the predicted value for a row can't ...) closed by gkronber
fixed: Fixed with r2393.
17:27 Changeset [2393] by gkronber

Fixed #765 (GP Predictor should output NaN if the predicted value for a row can't be calculated).

17:19 Ticket #766 (CEDMA server should start new runs of deterministic algorithms if the ...) closed by gkronber
fixed: Fixed with r2392.
17:15 Changeset [2392] by gkronber

Fixed #766 (CEDMA server should start new runs of deterministic algorithms if the settings for time-series algorithms are changed).

17:15 Ticket #766 (CEDMA server should start new runs of deterministic algorithms if the ...) created by gkronber
CEDMA server doesn't start new runs if the MinTimeOffset or …
17:13 Ticket #765 (GP Predictor should output NaN if the predicted value for a row can't ...) created by gkronber
At the moment an InvalidOperationException is thrown. This should be …
17:06 Ticket #764 (Delete CEDMA console) closed by gkronber
fixed: Fixed with r2391.
17:06 Changeset [2391] by gkronber

Fixed #764 (Delete CEDMA console).

16:51 Ticket #764 (Delete CEDMA console) created by gkronber
Has been replaced by new model analyzer.

09/25/09:

11:11 Changeset [2390] by mkommend

correctd bug in persistence of models (ticket #759)

09/24/09:

17:37 Changeset [2389] by mkommend

adapted modeling database (ticket #759)

16:52 Changeset [2388] by gkronber

Adapted ModelingResultCalculators to keep a mapping from ModelingResult to evaluators. Using mapping in DefaultOperators for data-modeling engines. #755

13:52 Ticket #763 (Plugin for data-modeling with k-Nearest Neighbour) created by gkronber
13:24 Changeset [2387] by mkommend

added GetModelingResultCalculator method to modeling result mapping (ticket #755)

13:09 Changeset [2386] by mkommend

added mapping between ModelTypes and ModelingResults (ticket #755)

11:37 Changeset [2385] by gkronber

Implemented #762 (ValidationQuality improvment progress as additional stopping criterion for default StructId Engines)

09:11 Ticket #762 (ValidationQuality improvment progress as additional stopping criterion ...) created by gkronber
09:10 Ticket #761 (Additional model quality metrics) closed by gkronber
fixed
09:09 Ticket #754 (The CEDMA controller should support setting the modeling type ...) closed by gkronber
fixed

09/23/09:

17:12 Changeset [2384] by mkommend

added possibility to create input variables in modeling database (ticket #759)

15:46 Changeset [2383] by mkommend

added mapping between ModelingResults and SimpleEvaluators (ticket #755)

14:05 Changeset [2382] by mkommend

refactored modeling database to enable the change of models (ticket #759)

13:58 Changeset [2381] by mkommend

added GetInputVariables to IPredictor (ticket #760)

13:38 Changeset [2380] by gkronber

renamed file MatrixCreator.cs -> Matrix.cs to reflect class name change in r2379. #761

13:35 Changeset [2379] by gkronber

Implemented additional model quality metrics. #761

11:23 Ticket #761 (Additional model quality metrics) created by gkronber
* general: NMSE (normalized Mean Squared Error) * time-series: (W)DS …

09/22/09:

17:10 Ticket #746 (CEDMA server is not compatible with new data-modeling algorithms) closed by gkronber
fixed
17:10 Ticket #757 (Predictors generated by SVM for time-series prognosis have to store ...) closed by gkronber
fixed
17:09 Ticket #750 (LinearRegressionOperator should first remove columns with many NaN ...) closed by gkronber
fixed
17:09 Ticket #753 (CEDMA Controller should adapt the training/validation/test set for ...) closed by gkronber
fixed: Fixed with r2375 (also see #754)
15:22 Ticket #756 (Enumerable for the algorithm type (Classification/Regression/Time-Series)) closed by mkommend
fixed
15:21 Ticket #749 (Move IteratePrefix of IFunctionTree from EvaluatorBase to new class ...) closed by mkommend
fixed
14:50 Ticket #731 (The modeling database backend should support a way to store ...) closed by mkommend
fixed
14:48 Ticket #760 (IPredictor should have a method to get all used variables) created by mkommend
IEnumerable<string> GetUsedVariables(); tbd with GK
13:47 Changeset [2378] by gkronber

Added "Set for all" button to simplify CEDMA algorithm configuration. #754

11:45 Changeset [2377] by gkronber
  • Implemented cloning and persistence in data-modeling algorithms.
  • Fixed bugs in CEDMA controller.

#754

09/21/09:

20:45 Changeset [2376] by gkronber

Fixed a small bug. #754 (The CEDMA controller should support setting the modeling type (classification/regression/time-series) for each target variable independently)

20:23 Changeset [2375] by gkronber

Refactored CEDMA dispatcher and CEDMA server view to allow different modeling scenarios for each variable. #754

15:39 Ticket #759 (The modeling database backend should provide methods to enable update ...) created by mkommend
15:07 Ticket #758 (CEDMA server should be based on new HL.MainForm) created by gkronber
15:03 Changeset [2374] by gkronber

Fixed a bug in extraction of variable impacts. #755 (Hard-coded strings for variable names of data-modeling results)

14:49 Changeset [2373] by gkronber

Fixed #757 (Predictors generated by SVM for time-series prognosis have to store min/max time offset to allow later application to new data).

14:49 Changeset [2372] by gkronber

Fixed a bug in extraction of TrainingMAPE. #755 (Hard-coded strings for variable names of data-modeling results)

14:15 Changeset [2371] by mkommend

added code for persisting of ModelType into the database (ticket #756)

09/18/09:

22:20 Changeset [2370] by gkronber

Added enumerable type for model results and changed analyzer model, the database backend and the model analysis operators to use the enumerable type instead of hard-coded strings. #755 (Hard-coded strings for variable names of data-modeling results)

20:27 Changeset [2369] by gkronber

Added enum ModelType (classification, regression, time-series-prognosis). #756

19:52 Ticket #757 (Predictors generated by SVM for time-series prognosis have to store ...) created by gkronber
19:49 Ticket #756 (Enumerable for the algorithm type (Classification/Regression/Time-Series)) created by gkronber
A field indicating the type of model has to be added to the model-type …
19:46 Ticket #755 (Hard-coded strings for variable names of data-modeling results) created by gkronber
The hard-coded strings should be replaced by an enumerable. The …
19:40 Ticket #754 (The CEDMA controller should support setting the modeling type ...) created by gkronber
19:38 Ticket #753 (CEDMA Controller should adapt the training/validation/test set for ...) created by gkronber
18:38 Ticket #752 (Plugin for data-modeling with decision trees) created by gkronber
18:37 Ticket #751 (Plugin for for data-modeling with ANN (integrated into CEDMA)) created by gkronber

09/17/09:

13:32 Changeset [2368] by gkronber

Fixed a bug in the variable impact calculation operators. #750.

10:15 Changeset [2367] by gkronber

Fixed #750 (LinearRegressionOperator should first remove columns with many NaN values and after that remove the rows with NaN values).

09:37 Ticket #750 (LinearRegressionOperator should first remove columns with many NaN ...) created by gkronber

09/16/09:

17:15 Changeset [2366] by gkronber

Fixed incompatibility between CEDMA server and time-series algorithms. #746 (CEDMA server is not compatible with new data-modeling algorithms)

17:14 Ticket #747 (GP engines create model with non-zero time-offsets) closed by gkronber
fixed: Fixed with r2365.
17:14 Changeset [2365] by gkronber

Fixed #747 (GP engines create model with non-zero time-offsets).

16:25 Ticket #741 (EvaluatedSolutions counter and current selection pressure is stored ...) closed by gkronber
fixed
14:26 Changeset [2364] by mkommend

added FunctionTreeIterator to HeuristicLab.GP.Interfaces (ticket #749)

14:18 Ticket #749 (Move IteratePrefix of IFunctionTree from EvaluatorBase to new class ...) created by mkommend
13:13 Changeset [2363] by gkronber

Fixed problems with classification algorithms. #746 (CEDMA server is not compatible with new data-modeling algorithms)

12:36 Ticket #748 (FunctionLibraryView is empty) created by gkronber
11:32 Ticket #747 (GP engines create model with non-zero time-offsets) created by gkronber
11:20 Ticket #746 (CEDMA server is not compatible with new data-modeling algorithms) created by gkronber

09/15/09:

18:16 Changeset [2362] by gkronber

Fixed #741 (EvaluatedSolutions counter and current selection pressure is stored incorrectly in models generated by GP algorithms).

16:38 Ticket #745 (More advanced linear regression methods with included feature selection) created by gkronber
16:37 Ticket #744 (Squared input variables in linear models) created by gkronber
16:36 Ticket #742 (Consistent naming scheme for data-modeling engines) closed by gkronber
fixed: Fixed with r2361.
16:36 Changeset [2361] by gkronber

Fixed #742 (Consistent naming scheme for data-modeling engines).

16:36 Ticket #706 (Linear regression algorithm for time series) closed by gkronber
fixed: Implemented with r2360.
16:35 Changeset [2360] by gkronber

Implemented #706 (Linear regression algorithm for time series).

16:02 Changeset [2359] by mkommend

changed behavior of bubblechart in ZoomToViewSize method (ticket #723)

16:00 Changeset [2358] by mkommend

set ActiveView to null during ViewClosed, if the active view is being closed (ticket #743)

15:59 Ticket #743 (Views are not correctly disposed during ViewClosed) created by mkommend
14:59 Ticket #742 (Consistent naming scheme for data-modeling engines) created by gkronber
[…]
14:50 Ticket #741 (EvaluatedSolutions counter and current selection pressure is stored ...) created by gkronber
14:48 Ticket #705 (SVM for time series prognosis) closed by gkronber
fixed
14:48 Ticket #625 (Algorithm for support vector classification) closed by gkronber
fixed
14:48 Ticket #740 (SimpleEvaluators in HL.GP.StructId and HL.SVM are not compatible with ...) closed by gkronber
fixed
14:47 Ticket #739 (Engine for classification based on linear regression) closed by gkronber
fixed
14:07 Changeset [2357] by gkronber

Fixed #740 (SimpleEvaluators in HL.GP.StructId and HL.SVM are not compatible with evaluators in HL.Modeling).

14:06 Ticket #740 (SimpleEvaluators in HL.GP.StructId and HL.SVM are not compatible with ...) created by gkronber
In HL.GP.StructId the output matrix contains the original value …
14:01 Changeset [2356] by gkronber

Moved common (not algorithm specific) model analyzer code (same operator tree in all regression/classification/time-series-prognosis engines) to HL.Modeling. #625, #705, #739.

10:47 Ticket #726 (Dataset should provide method to get all values from a variable) closed by mkommend
fixed
09:48 Ticket #732 (Observable Collection is needed for the ModelAnalyzer) closed by mkommend
fixed
09:48 Ticket #716 (Prototype implementation for generic main form) closed by mkommend
fixed
09:47 Changeset [2355] by mkommend

stored model metadata into modeling.database
changed analyzermodel to use dictionary of string, object (ticket #731)

09:46 Ticket #712 (Redesign database backend for HeuristicLab.Modeling) closed by mkommend
fixed

09/14/09:

18:48 Changeset [2354] by gkronber

Added RandomInjector (necessary for selection operator) to TimeSeriesSupportVectorRegression engine. #705

18:48 Ticket #736 (AnalyzerModel should also support time-series or classification ...) closed by gkronber
fixed
18:43 Changeset [2353] by gkronber

Added default engine for linear regression based classification. #739

18:09 Ticket #739 (Engine for classification based on linear regression) created by gkronber
17:35 Changeset [2352] by gkronber

Using simple version of TheilInequalityCoefficientEvaluator in both, SVR and GP (for time-series).
Deleted GP-specific version of TheilInequalityCoefficientEvaluator. #705

17:30 Changeset [2351] by gkronber

Worked on SVM classification engine. #625

14:27 Changeset [2350] by mkommend

corrected usage of new and override keywords for inherited properties from control or form in MainFormBase (ticket #716)

12:59 Changeset [2349] by gkronber

Created a subclass for time-series prognosis with SVM and created a simple version of TheilInequalityCoefficientEvaluator in HL.Modeling. (#705)

10:25 Changeset [2348] by mkommend

adapted mainform base to declare gui specific members with the new keyword (ticket #716)

10:08 Changeset [2347] by gkronber

Worked on SVR algorithm for time-series. #705

Note: See TracTimeline for information about the timeline view.