Chameleon Home Seventh Sense Software Chameleon Homepage Features News and Updates Developer's Toolkit Reference Materials Download Trial Version Support and Feedback Image Gallery

RELEASE NOTES

Below are release notes for earlier versions of Chameleon Statistics Scientific Edition. (Release notes for latest version available here).

Version 1.2 - New features:

* Density estimates can now be exported as human-readable Chameleon data files.
* Additional data normalisation options added for improved classification accuracy.
* Hourglass and countdown implemented for large classification tasks.
* Class symbol functionality improved. Default symbol option added.
* Image data samples are now labeled by image coordinates for easier reference.
* Cluster size and fusion heights are now displayed in scatter plot and binary tree.
* Tutorial section added to help manual.

Bug fixes:

* Rotation options problem fixed (v1.22 only - earlier release contains corrupt data)
* Data editor improved to allow access to and editing of all data at all times.
* Incorrect sample labeling error in editor for unlabeled datasets has been fixed.
* Problem displaying spanning tree in the presence of subclusters has been fixed.
* Class colours are now displayed correctly after loading Chameleon project files with hierarchically clustered data.

Known Issues:

* Problems are reported to occur occasionally during import of manually generated data sets. (Workaround: Make sure text data files end with a new line character or an empty line. Export data sets in a standard Chameleon format once the data has been imported correctly).

Version 1.1 - New features:

* Explicit manual entry of binary tree cut height has been implemented.
* Data values can now be viewed and edited at all times. The editor now displays sample and variable labels rather than numerals for easier reference.
*
Contours can now be overlayed onto plain and density maps.
*
Unique symbols are now available for each class. A triangle symbol has been added.

Bug fixes:

* The sample labels sometimes began with Sample 0 instead of Sample 1 in the profile and scatter plot views. This has now been corrected.
* Performing single-link clustering on partitioned data when classes with only a single member exist caused the program to crash. This has been fixed.
* Spanning tree connections joining samples belonging to different classes are no longer displayed.
* Class colours are now displayed correctly after non-single-link hierarchical clustering. An hourglass and clusters countdown has also been added.

Known Issues:

* Problems are reported to occur occasionally during import of manually generated data sets. (Workaround: Export data sets in a standard Chameleon format once the data has been imported correctly).