 |
|
Changelog
Version 2.3.1 (2009/09/03)
Major changes:
* Visible interaction between predicted DataSets and models used for prediction.
* Pixels with no variation is automatically excluded when importing or predicting image data.
* New feature for setting number of model components without having to adding/removing components.
* New functionality for merging several images (tif, png,jpeg etc.) by layer into a single image.
* Score and loading plots now show the explained variance of the plotted components.
* Fixed bug regarding saved models losing information about identifiers and classes.
* New overview plot for model containers, which show explained variances for all included models at the same time.
* The Evince default colors have been changed for plots and data tree items in order to increase perception.
* Manual selection of dcrit for SIMCA classification.
Minor changes:
* Fixed so that full length of object names is visible in data tree.
* Bi-plot axis for both layers is synced when changing component.
* Correct scaling of score plot axis with hotelling ellipse.
* Saving project upon closing down Evince now saves project into recent project list.
* Fixed crash with reduce to average transformation when the total number of observations was divisible with the reduce to average group size.
* RMSEP table is updated correctly when the number of PLS components is changed.
* Increased performance of spectral plot.
Version 2.3 (2009/06/08)
Major changes:
* It is now possible to get reflectance values higher than one unit by unticking "Cap reflectance values higher than one unit" during import.
* Better performance when calculating models, predictions and transformations due to better use of multithreading.
* Possible to predict datasets with a different number of variables than what is present in the model.
* Possible to save spectra from spectral plot to file.
Minor changes:
* Added Leave-One-Out Cross-Validation
* Fixed crash when predicting external data from single-Y PLS models.
* Fixed crashed when creating histogram for ycalc.
* Fixed interaction for Bi-plot.
* Average for selections in quick information for Contour 2D plots.
* Minimum and maximum values corrected for legends.
Version 2.2.2 (2009/02/24)
Major changes:
* Added Bi-plot that shows normalized T values together with P values in the same plot
* Printing functionality is available for plots
Minor changes:
* Fixed problem with saving after removing models
* Fixed axis ticks for contour 2D and column plots
Version 2.2.1 (2009/02/04)
Major changes:
* Memory optimizations. Imported data can now be stored in short or single
precision in order to preserve memory.
* Added SIMCA Classification
* Added Cooman's Plot
* Fixed problem with adding components to PLS single-Y models
* Updated help section to reflect recent changes in functionality
Minor changes:
* Fixed problem with opening SAC-files with special characters in the name
* Fixed problem when exporting to sdf from PCA class models
* Improved MATLAB export
Version 2.2.0 (2008/11/14)
Major changes:
* Support for import of multiple files. New tab selection in the import
wizard "Multiple image import", for merging multiple files into on project.
* Multiple legends in the plot properties dialog.
* The user can now change the order of classes inside legends by moving
classes up or down in the class settings panel. The class settings panel
is found in the data set table.
NOTE: the sort option in the legend properties must be unchecked.
* Added possibility to save models containing only the necessary data needed
for an external prediction.
Minor changes:
* Now possible to add more then one wild card for string selection in the
selection panel, in data set tables.
* Legends can show items by class from both color and shape.
* Improved script functionality.
Version 2.1.0 (2008/10/06)
Major changes:
* Missing values can now be replaced by median value of surrounding pixels
during import.
* A new modification, Prediction classes, added. Automatically created by
the PLS-DA model. This enables automatic classification from the classes
specified in the model. This can be used to color plots, create layers
and/or select data points. More information can be found in the help
shortly.
* When creating a PCA model from classes, the models will be stored in a
model container that can be saved as a single model. It will also be
updated automatically if the data set is changed.
Minor changes:
* Tweaked MATLAB import and export.
* Now possible to drag pixels from RGB plots to contribution plot settings.
* Tweaked information shown in tooltips for some plots.
* Improved X-axis descriptions for Spectral plots; this means that descriptions
will now be fetched from the last data set interacted with instead of always
from the original data.
* Supports the SAC file format (Legacy and standard).
* Optimized calculation speed of predictions and transformations.
* Improved script functionality.
* Updated 3D library to latest version.
Bugs fixed:
* Colors should no longer reset in plots belonging to a different data set
when changing the current data set.
* Pasting of values during import should work better; rows would sometimes be
transposed.
* The user will get a warning while trying to create a PLS-DA model if here
is no variance in the Y variables. Earlier this lead to an error message.
* Fixed problems with moving classes to layers.
Known bugs:
* Problems with 3D plots under Linux.
* No support for 3D plots in 64bit Java under OSX.
* Some problems under Windows Vista when computer is in a domain.
Version 2.0.21 (2008/08/07)
Major changes:
* Interactions has been overhauled
* Read support for SAC (SisuCHEMA Archive Container) file format.
Minor changes:
* Wild card search for variables has been added.
* Logarithm transformation can now either update values or add new values
(old default was new value).
* Reverse colors for plots now possible.
Version 2.0.20 (2008/05/28)
Major changes:
* MATLAB export greatly improved.
Minor changes:
* Help chapters updated to reflect changes made.
* Enabled removal or renaming of identifier columns.
NOTE: For changes to be reflected in the models, changes must be applied
* Increased speed of copying plots to clipboard, especially for some plot types.
Version 2.0.19 (2008/05/14)
Major changes:
* Plot legends have been improved.
* Modifications in the data set are now shown in the order they are calculated.
* Logarithms can now have constants (log(X + C))
* When creating classes, wildcards can now be used.
* Wildcard selection for observations in the data set.
Minor changes:
* Screen shots shown in preview for new saved projects.
* Excel import times improved when using file with several sheets.
* Script improvements.
Version 2.0.18 (2008/04/22)
Major changes:
* New functionality for manually adding Y variables by setting a selection to
a specific value.
* Improved image import where a preview of the image is shown. It is possible
to exclude both pixels and wavelengths in the image preview.
* Added functionality for correcting images with a set of standard images.
Spectral unit conversion to reflectance and pseudo-absorbance is possible.
* Support for reading .HIPS file format.
* Support for writing MATLAB files (.mat) and ENVI files.
Minor changes:
* Observed vs Calculated plot for PLS models is easier to use.
* Observations/variables from a test set is greyed out when excluded.
* Added warning when it is likely too many subclasses would be created from
some actions (would be very slow).
* New (not loaded) Spectral plots from original data will have a set
maximum value (based on maximum in the data) .
* Model collections have been tweaked to display more useful plots.
* All functionality in the data set menu for observations are always
available in order to avoid confusion.
Version 2.0.17 (2008/03/04)
Major changes:
* Added support for 3D hardware acceleration for Contour 3D plots.
NOTE: 3D acceleration can be turned off in Tools/Options/Global 3D
Settings.
* Add/remove component functionality in model overview plot
* Logarithm transformation is fixed when missing values are present
* Added support for creating spectral angle maps from image data
* Data range for Contour2D, Contour3D and RGB image plots is now updated
when new component is added or removed
* The viewing of identifiers is improved.
* Fixed error for RGB image plots when certain observations are excluded.
 A modern software for multivariate data analysis & chemometrics  |
 A complete system for keeping track of your chemical storage |
 Multivariate data analysis consulting within biotechnology and chemistry |
UmBio AB, Box 7980, 907 19 Umeå, Sweden, Phone: +46 (0)90 - 15 48 20, Fax: +46 (0)90 - 15 48 67, info@umbio.com |
||