search

henryzord / ardennes

An Estimation of Distribution Algorithm for Decision-Tree Induction.
5 stars 1 forks source link

Twin nodes #24

Closed henryzord closed 7 years ago

henryzord commented 7 years ago

One way of solving the problem of splitting categorical attributes (and thus preventing the tree from losing too much height in dealing with multiple splits) is to perform a n-ary split (where n is the number of categories for a given predictive attribute), but set all children nodes with the same node label.

Obviously it is required to decrease the impact of this repeating of attributes in the GM update.

henryzord commented 7 years ago

Will change the GM to use only one variable per level. This should resolve this issue as well.