has been test for all methds
16/3/24
added categorise_num: for two-stage regressionly only
default is supervised aka retain y when splitting
splitting rule on max_level = 2,
idea was to use internode variance - but waiting for advise on this
only implemented at start of tree growing, with flexibility to expand to each father node
timign takes a little bit , for the most part implemented with numpy
can see eta^2 is nto close to 1 for num vars now.
5/6/24
Pypi uploads required a change in name of the library from TREEplus to TREE4.
TREE4 can be pip installed
And teh objects in the library activated by TREE4.TREE4 import *
Also uploaded the reference manual
Made a change to the font size of hover text for readability
made a change to lba fucntion only allowing more than 1 row item for the cross tables passed to lba function