Updated documentation

v0.7.5

Dataset preview and result display in user GUI windows

This release provides a richer feedback to the end user. First, once a project data set is chosen, it can be "previewed" in a modeless dialog window. When the application is run, results are not only echoed to the console but also echoed to the end user in windows.

Parameters interface tab managed more sensibly in between projects.

Internal refactoring to allow better management of parameter page (tab) presented to the user.

Workbench autoloads properly upon package installation

v0.7.2

autoloading (with correct logo display)

Implementation of some measures for missing data (NA) management

This release implements two functions to manage missing data in input trait data sets:

1. columns and rows completely filled with missing data (?!?) are filtered out during data project creation.

2. Missing data is detected during "Check Parameters" feature of the parameter setting step. Users may (optionally) run the Amelia package imputation process on the data set to infer in missing data. The initial implementation is experimental and doesn't (yet) leverage other parameters to guide the processing (e.g. Amelia ought to be told what variables are continuous versus nominal versus ordinal, but not yet done... Other tweaking of the procedure may also help in the future...).

Iterations on Alpha2 based on project team feedback

• Multiple "Set" buttons in parameter screen replaced by a single "Check Parameters" button for all changeable parameters. Original "Set" didn't really set but only validated the parameters. Check Parameters either reports "OK" or give specific error message.
• Trait filter threshold values may be left blank in which case, the absolutely observed upper (max) or lower (min) values are used as the threshold value.
• Logo cleaned up a bit.

Tagged analysis runs and automated deferred sample distribution generation

This release has two Explora enhancements:

1. Analysis runs are now tagged (default: an automatic time stamp is used) and the data tag used to partition analysis runs into their own subdirectory in the main "project" subdirectory, allowing for repeated (and variable) analysis runs.

2. The need for users to explicitly generate the input sample distribution for multi-objective selection has been removed by automatically generating the distribution the first time an accession selection algorithm is run (then caching the resulting distribution for reuse by other available selection algorithms).

Minor enhancements to the user interface text was undertaken to clarify the intent of the tool operations.

Minor bug fix to v0.4.0

Decision tree graphics didn't display correctly without an additional library import.

4th Alpha release of Explora

A significant amount of debugging and remedial repair of the code base appears to have rendered functional most (if not all) of the existing optimization algorithms. When given sensible input values, these algorithms now appear to give sensible outputs, including the expected graphics (for PCA and Decision Tree runs).

This version should be extensively tested by project members who know the scientific nature of the tool and understand what kinds of outputs should be generated given a specified set of sample inputs.

3rd iteration on Alpha2

Software resets itself whenever a new dataset is selected from the Project dropdown box (view tabs also get reset...)