Class aggr_adaboost: is there a way to make this testable as classifiers are added?
Class aggr_hiesplit: think about the future of this class, it cannot be trained like the others! The problem is that the classes in dataset passed for training are not the classes classified by the classifier
Class as_fsel_eclass: Temporarily deactivated
Member block_cascade_base::do_train (in o, in data): think about stacked generalization
File confusion_str.m: no longer managing rejected properly
Class fcon_spline: Should make the basis on training, not only use!
File format_number.m: Introduce setup for this function. E.g. Nature requires comma separators.
File get_credits.m: mention another publication
File get_suffix.m: Could respond to IRootLab setup to change the way new generated objects are named, i.e. how they appear in objtool and datatool
File params2str.m: find and group documentation about this parameters convention
Class report_soitem_sovalues: implement the params GUI
Class reptt_bag: temporarily deactivated.
Class reptt_hiesplit: Temporaritly deactivated. This may be obsolete. There is new dataset property instead of this hie_split/hie_classify pair
Class vis_as_fsel_hist_entropy: this is obsolete. Should take log_hist as input, not as_fsel_hist (which doesn't exist anymore)
Member vis_clssr2d::do_use (in o, in obj): still TODO
Class vis_log_celldata: needs implementing the properties GUI