Gradient chart view for regression solutions

[bburlacu]

As the name suggests, this view contains a chart plotting the gradient of the target variable with respect to a chosen input variable. The user may then vary the input value (by means of mouse interaction with the chart) in order to see how the target response changes.

The gradient chart supports synchronisation of backing data using a shared modifiable dataset, allowing the user to get a complete view of the target variation by arranging side by side multiple charts corresponding to different input variables.

Additional features:

• Responses for variables that are not used in the model should be grayed-out or hidden. Thos, methods for querying the actual used inputs for a regression model are implemented.
• If a model can estimate the variance of a sample (e.g. Gaussian process), the confidence interval should be shown.

Example gradient chart:

[bburlacu]

r13780: Initial commit implementing gradient view and chart.

[bburlacu]

r13783: Fix text annotations.

[bburlacu]

r13808: Start implementing table layout panel for gradient charts in a separate view.

[pfleck]

r13810: Branched HeuristicLab.Algorithms.DataAnalysis.Views plugin. r13811: Updated project references.

[pfleck]

r13812: Added an new View for GaussianProcessSolutions to be able to interactively show the gradients and confidence intervals for user-input ranges.

[pfleck]

r13816:

• Fixed typo in class/filename.
• Added new control that shows density information adjacent to the trackbar.

[bburlacu]

r13817: Added RegressionSolutionTargetResponseGradientView and updated GradientChart.

[bburlacu]

r13818: Fixed small bug when the values are not correctly rendered in the chart initially. removed unused variable limits from view.

[pfleck]

r13819 Branched HeuristicLab.Algorithms.DataAnalysis and HeuristicLab.Problems.DataAnalysis.