ciu 0.8

• Changed the default colours if influence/contrastive plots to the brick red / blue used by LIME in R version.
• Corrected the function “ciu.contrastive” that was implemented in a very strange (and presumably incorrect) way.
• Added possibility to get line segment IO-plots for categorical features, rather than the earlier histogram visualisation. Added parameter to ggplot.ciu() “categorical_style=”segment”“, where the default is now to use line segments, even though that breaks backwards compatibility a little. However, for some reason setting”ylim” breaks the old “geom_col” plot so the bar plots are now un-scaled. But, strange enough, “+ ylim” applied to the returned ggplot works so it’s possible to adjust afterwards. Something happened somewhere, no change in the CIU code and it worked in the past.
• Added new function ciu.contrastive that uses CU values of one class/instance/whatever as the baseline “normal.CU” values of the other.
• Added new function ciu.contextual.influence that calculates contextual influence value(s) from CI and CU value(s) and “baseline”.
• Added new file ContextualInfluence.R for functions related to contextual influence.
• Fixed x-axis label error for influence plot in ciu.ggplot.col.

ciu 0.6.0

• Extensive tests on Intermediate Concepts, which are also integrated into the new README.Rmd file (and README) as well as in test scripts TestCases.R and TestCases_NoObject.R on Github.
• Fixed inverted axis bug in plot in plot.ciu.3D.
• Integrated support for mlr3. Only tested with “classif.rpart” model though because other tested models (xgboost, nnet, …) don’t like R types and no need/time to dive into mlr3 for the moment. To fix in the future in case bugs are discovered.
• Added parameter “plot.mode” to “ciu.ggplot.col” function, which allows for value “overlap” that plots CI as a bar and then CU as another bar over it, scaled into CI bar. Colour of CU bar can either be fixed (dark green by default) or set to NULL, which then uses the green-yellow-red color scale (unless those colours are changed by the function parameters).
• Removed parameter for min/max values of contextual influence because the range of contextual influence is (and should be) always one (1) based on the mathematical constructs.
• Added new method “ggplot.ciu” to ciu.new for plotting input/output graphs with ggplot. Identical to old plot.ciu, except that 1) ggplot offers some advantages, notably what comes to figure scaling, 2) added possibility to include CIU visualisation (cmin, cmax, neutral) by setting “illustrate.CIU=TRUE”.
• Added new method “influence” to ciu.new for getting numerical contextual influence (rather than just seeing them in plots). TO-DO: add corresponding function to “ciu.R”.
• Corrected axis in plot.ciu.3D.

ciu 0.5.0

• Textual explanations have been implemented with function “ciu.textual”.
• Implemented “meta.explain” method, which returns a self-contained “ciu.meta.result” object with CIU results for a given instance. This mainly makes it possible to visualize exactly the same CIU result in different ways. Before this (and still if no “ciu.meta” parameter is given), every visualization method ran “explain” again, so CIU results could differ somewhat between different runs.
• Added parameters “use.influence” and “influence.minmax” to “ggplot.col” and “barplot.ciu” functions/methods, which produces a LIME/SHAP/etc-like plot where “influence = (influence.max - influence.min)*ci(cu-neutral.CU)“, bars go either right or left of zero and there are only two colours.
• Added support for factor-type inputs to method “$plot.ciu”.
• Created new “class” called “ciu”, which is just a list object with the “instance variables” of a CIU object. This makes it possible to create plot functions that are not methods of CIU but rather take a “ciu” object as their first parameter. The main reason for doing this modification is that CIU objects seem to use much more memory than a simple data-based “ciu” object. The documentation and package tools in R are also not too aware about “inner functions” (methods), which is an inconvenient. Finally, the plan was indeed to go for this approach in any case because increasing the code length of CIU was not desirable in the long run. The way in which it is implemented now allows functions and methods to be used interchangeably in any case.

ciu 0.1.0

• First version of ciu published at CRAN