Laboratoire ICAR » Plateforme TXM

DONE: separate UI code from data and computing code in CAHEditor, CAHEditorInput and ComputeCAH classes
DONE: add some methods in CAH to provide better access and cache to and from the R FactoMineR result as in CA class (e.g. getClusterCoords(), getHeights())
DONE: the CAH command uses now the SWT chart components provider (but R charts engine usage is forced even if the current chart engine is JFC)
The CAH 2D JFC chart implementation is still in progress.

DEV important notes:
The FactoMineR R HCPC aggregation contains an nrow(parts) * 2 matrix with the merge data of clusters in res\$call\$t\$tree\$merge, for example:

-5    -6
-3    -4
-7    1
2    3
4    5

The absolute number values are the index of rows starting at 1 (in the unsorted row names vector). Negative value means a pair.
The heights are stored in res\$call\$t\$tree\$height.

The HCPC plot f.draw.tree sub-function does not seem to use the heights for drawing the tree but seems to directly use the coordinates.

The eigenvalues are stored in res\$call\$t\$res\$eig.

At this moment I'm trying to port the R source code of f.draw.tree to Java.