Timeline
09/21/16:
- 21:27 Changeset [14300] by
-
#2675: fixed TimeframeFeatureCorrelationCalculator
- 21:21 Changeset [14299] by
-
#2674: also added row names to clipboard for StringConvertibleArrays (the code for copying to clipboard was irrelevant anyway because C&P was not disabled for the DataViewGrid)
- 21:19 Ticket #2675 (TimeframeFeatureCorrelationCalculator cuts the observations within the ...) created by
- The expected behavior is that the calculator determines correlation …
- 19:58 Ticket #2674 (Copy&paste should also copy the row names for DoubleArrays) created by
- 19:56 Ticket #2673 (Solution view for variable impacts) created by
- The regression solution view has a separate button to calculate and …
- 10:13 Changeset [14298] by
-
#2661: catch and handle all Exceptions because otherwise they are swallowed by the TPL
- 09:55 Changeset [14297] by
-
#2667: catch all InvalidOperationExceptions (ObjectDisposedExceptions) for the invoke in ProgressView
- 09:49 Changeset [14296] by
-
#2661: implemented fixes for several problems in the TableFileParser. We now also store the original string representation of all tokens and use those when we detect that a column cannot be read as DateTime / double.
09/20/16:
- 17:34 Changeset [14295] by
-
#2672: Fix typo bug in cloning constructor.
- 17:14 Changeset [14294] by
-
#2672: Add cloning constructors.
- 15:51 Changeset [14293] by
-
#2672: Implement online calculator cloning.
- 15:50 Ticket #2672 (Online calculators should be cloneable) created by
- In certain use cases it would be useful for performance reasons to …
- 14:32 Ticket #2671 (It should be possible to view / change random seeds for the problem ...) created by
- See #2371
09/19/16:
- 17:54 Changeset [14292] by
-
#2649: Corrected small mistake in the OnlineMeanAndVarianceCalculator wrt the various error states it provides.
- 17:46 Changeset [14291] by
-
#2660: changed calculation of variable relevance values for variable interaction networks based on sampling Gaussian processes.
Instead of taking inverse length scale (ARD). We calculate the deviation from the original function (y) after individual variables are removed (y').
- 15:27 Changeset [14290] by
-
#2669: Added setter for the TargetVariable property in the classification and regression model interfaces and adjusted implementing classes accordingly. Similarly, added a TargetVariableChanged event to the models.
- 15:15 Ticket #2670 (Target Response Gradient charts should be renamed) created by
- The charts do not actually show gradient information. We still need …
- 11:23 Changeset [14289] by
-
#2669: Removed the after deserialization method that resets the targetVariable. Instead the initialization code has been added to the storable ctor.
- 11:21 Ticket #2669 (TargetVariable information is lost in DataAnalysisModels) created by
- With r13992 the target variable has been set to string.Empty. This has …
09/17/16:
- 19:01 Changeset [14288] by
-
#2661: fixed build fail in HeuristicLab.Tests
- 18:59 Changeset [14287] by
-
#2667: implemented a workaround objectdisposedexception crashing HL. Hard to debug.
- 18:37 Changeset [14286] by
-
#2661: make sure the progressbar is removed when the TableFileParser throws an exception
- 18:34 Changeset [14285] by
-
#2661: make sure progressbar is removed when the TableFileParser throws an exception
- 11:24 Changeset [14284] by
-
#2661: added unit tests to reproduce the described bugs
09/16/16:
- 15:36 Changeset [14283] by
-
#2288: Small refactor
09/15/16:
- 15:15 Changeset [14282] by
-
#2668: Initialized interpreter field in the LinearExpressionTreeInterpreter deserialization ctor.
- 10:16 Changeset [14281] by
-
#2635: Small cosmetic refactor.
09/14/16:
- 23:04 Ticket #2668 (Private field 'interpreter' is not initialized after deserialization ...) created by
- This leads to exceptions when using e.g. integral symbols which falls …
- 16:56 Changeset [14280] by
-
#2635: Small refactor.
09/13/16:
- 16:43 Changeset [14279] by
-
#2635: Implement an alternative way of assessing which offspring should be rejected early: the child is evaluated on part of the training data and the remaining rows are assumed to be perfectly correlated with the target. If the offspring selection criteria is not fulfilled, we can reject early. Otherwise the child is evaluated on an additional slice of the training and so on.
09/12/16:
- 12:01 Changeset [14278] by
-
Initial Commit for ticket #2605
Implemented:
- Encoding
- Moves
- Views
- Blocks World Problem
- Blocks Relocation Problem
- Stacking Problem
09/08/16:
- 15:58 Ticket #2667 (ObjectDisposedException might crash HL in ProgressView in combination ...) created by
- I got an exception in the following code when accessing progressBar …
- 11:41 Changeset [14277] by
-
#2650: merged r14245:14273 from trunk to branch (fixing conflicts in RegressionSolutionTargetResponseGradientView)
- 11:25 Changeset [14276] by
09/07/16:
- 15:20 Changeset [14275] by
-
#2288: Clean up code and add comments in the ConstrainedForceDirectedLayout class. Minor changes to view and directed graph chart. Introduced an INetworkNode interface for more flexibility. Updated cola and adaptagrams dlls with latest changes from upstream.
- 13:52 Ticket #2666 (Improve Performance of LinearLinkageEncoding) created by
- LinearLinkage.FromLLEe uses a Dictionary<int,int>. Replacing that with …
- 10:31 Changeset [14274] by
-
#2665: Added folder for PushGP branch.
- 10:30 Ticket #2665 (Implement PushGP) created by
