Hi Erica,HHEd has "TR" command to control log levels. Here are log levels related to clustering.#define T_CLUSTERS 0x0100 /* Show contents of clusters */#define T_QST 0x0200 /* Show items specified by questions */#define T_TREE_ANS 0x0400 /* Trace best unseen triphone tree filtering */#define T_TREE_BESTQ 0x0800 /* Trace best question for splitting each node */#define T_TREE_BESTM 0x1000 /* Trace best merge of terminal nodes */#define T_TREE_OKQ 0x2000 /* Trace all questions exceeding threshold */#define T_TREE_ALLQ 0x4000 /* Trace all questions */#define T_TREE_ALLM 0x8000 /* Trace all possible merges */So if you runTR 2049(2049 = 0x0801) just before TB commands, (hopefully) HTS will output both normal log plus best questions at each node. Likewise, if you turn on T_TREE_ALL_Q, it will generate log for all questions. You can try other values or just comment-out related lines in the HHEd.c:ValidProbNode() to generate this information.Cheers,HeigaOn Wed, Mar 8, 2017 at 8:48 PM Erica Cooper <ecooper@xxxxxxxxxxxxxxx> wrote:Hi all,Is there a way to find out how informative a given question was in the decision tree construction, e.g. by finding out what was the likelihood gain at a given node split? I haven't found anything like this in the training log output or in the tree files themselves, but maybe this information is available elsewhere in the training output? or would I have to modify the code to log this and re-train?Thanks,
Erica--------------------------------Heiga ZEN (in Japanese)---------
Byung Ha CHUN (in Korean)
<heigazen@xxxxxxxxxx>