Forest-based Classification and Regression Tool - prediction mode

[RobertFraser]

Does anyone know if there is a way to train a random forest model and then be able to apply it to several different datasets using this tool? It appears that the way the tool is designed, it wants to train a new random forest each time you run it, even in "Predict to raster" mode. Is there a way to lock or save a RF model and then apply it several times to different datasets?

Thanks.

Rob

[RobertFraser]

OK - think I figured this out. If you want model stability amongst training runs you need to change the default Seed value for the random number generator under the tool's Environments settings. I changed the seed to value of 3 and then the results from consecutive training runs were identical.