Release Notes Version 8 7 1

From Eigenvector Research Documentation Wiki
Jump to navigation Jump to search

Changes and Bug Fixes in Version 8.7.1

This release includes some updates for Matlab 2019b. The software will still show a warning that can be ignored for 2019b.


  • jcampreadr - Add support tabular data form = "##XYPOINTS", in addition to "##XYDATA" and "##PEAKTABLE".
  • opusreadr - Changes to handle Bruker Opus files having data and parameter block types with bit 31 = 0 indicating the file was modified by the user. Setting this bit to = 0 is a 21CFR part11 package feature. The block type's bit 31 is = 1 when the file is created by the instrument. On import, the block with the requested spectrum type having bit 31 = 0 is read. If this does not exist then the block with the requested spectrum type = 1 is read. This is done for data and parameter blocks. It is assumed there is one measured sample per file.
  • rdareadr - Add reader for importing Siemens .rda file into a DataSet Object.
  • visionairxmlreadr - Major performance improvements.

Additional Changes

File Comment
  • Add support for 3rd party features.
  • Fix bug preventing prediction when using only 2 parameters, model and data, (but no options).
  • Enable logical array or a numeric array of 1's and 0's as indices.
  • Added compression, maxpcs = 40, updated algorithm.
  • Changed flucut so it doesn't soft delete the blank sample.
  • Convert more interfaces to use this as default.
  • Change default to use values returned by Matlab (rather than Java). Help improve appearance on high DPI screens.
  • Fix Scores Plot's calculation of Q and T2 "reduced" which was calculated incorrectly by assuming the model was built using options.confidencelimit = 0.95.
  • Add code to restore colormap if plot command changes it.
  • Extend to support negative input T-statistics (in addition to positive). Now similar to normdf.m.
  • Add the jar needed for XGB on Linux.
  • Ensure XGBDA works properly if test data has no classes (or y-block).
  • Fix bug when number of supplied filenames equals number of included samples, which could cause samples to be written to the wrong files.