09-16-2015 11:10 AM
My team has used Enterprise Miner 13.2 to build out several decision trees. A few of these trees were built interatively and, now they're finalized, we want the tree structures to remain static.
We've been setting the "Use Frozen Tree" option to "Yes" to keep the tree structures in place.
We've recently received a fresh input dataset which has a different number of rows than what was used to develop the static trees. We would like to see how/if this fresh input dataset affects results. However, when we re-run the frozen decision trees, they seem to not ignore the new observations. If we reset the "Use Frozen Tree" option to "No" and re-run, we have to manually rebuild the trees interactively. Periodically refreshed datasets are unavoidable and we'd like to avoid having to manually rebuild trees every time.
Is there a way to preserve a tree's structure, but simply run new data through it?
Thanks in advance,