search

starling-lab / BoostSRL

BoostSRL: "Boosting for Statistical Relational Learning." A gradient-boosting based approach for learning different types of SRL models.
https://starling.utdallas.edu
GNU General Public License v3.0
32 stars 21 forks source link

Discretization of Continuous-Valued Attributes #29

Open harshakokel opened 4 years ago

harshakokel commented 4 years ago

The discretization code currently does independent discretization of test and train set. This is a problem when the test and train sets do not follow same distribution.

Ideally, same bins should be used for both the sets. Saving the bins used while discretization of train set and using the same bins while discretizing the test set could be one solution.