GitLab

Change-Id: I24459365cc97feabf8cecb8baa95c8617c38b9e3

NB: Too small samples can generate test errors, for the computed
results are tested statistically.

re-declarations shadowed previous declaration

ENH: complete overhaul of the statistics algorithms so that models are now vtkMultiBlockDataSets only
ENH: updated tests accordingly

ENH: modified descriptive statistics so that model port is now a multiblockdataset (and no longer a table). This has many advantages, in particular (1) in terms of consistency with more recent engines (2) to establish a clear distinction between primary and derived models and (3) to allow for subsequent development of additional derived statistics
ENH: modified test harness accordingly

     LearnOption, DeriveOption, AssessOption, in order to avoid confusion
     with the fact of performing the corresponding calculations
ENH: updated all tests accordingly

     the potentially unlimited size of the test sample. NB: with 6 standard deviations
     (and even with 5), it is to be expected to find that 100% of the sample points
     are within range (i.e., no outliers) if one runs with less than several dozens of processes.

     respect to correlative and descriptive statistics

ENH: updated tests accordingly
BUG: removed a source of memory leak

     multi-correlative engines
ENH: updated serial and // tests accordingly

     vs. the hand-calculated aggregation of all serial statistics
     (descriptive only for now).
ENH: print extrema info also (descriptive only)

     controller)

     and subsequently chooses one.

ENH:  Implemented Parallel PCA Statistics.
ENH:  Added the test for Parallel PCA Statistics to TestParallelRandomStatistics.cxx
ENH:  Added Parallel PCA Statistics to the build.

ENH: by default, test more demanding data sets

     variable parameters, and reporting error codes).

ENH: added test for the above.

       Mahalanobis distance in 1D.
ENH: updated tests accordingly