we're currently trying to train a decision tree lerner knode on a data set of 62000 rows and about 150 columns. It took about an hour to finish 1% of the process. Is there any possibility to excelerate the process by changing the settings of the knode or sth. like that?
Hi, please have a look into the node dialog (configure). There is a field called "Min number of records per node", if you increase this number the Decision Tree Learner should speed-up; also if you have many nominal values, you might increase the "Min. nominals" to something like 100 or so. For more details on the node options, please have look to the Node Description. Regards, Thomas