In SPM v7.0 (Salford Systems' latest product release) one of the new features available in the TreeNet component (Ultra version) is the ability to build Random Forests models. By building the model in the TreeNet engine, you're able to take advantage of the powerful technology offered by gradient-boosted trees.
The main advantages of using TreeNet (stochastic gradient boosting) to build Random Forests models are:
- Multithreading capabilities
- Partial dependency plots
- Speed of model building
- Interaction detection
To help explain how to do this Salford Systems' Senior Scientist Mikhail Golovnya has created a short video demonstration that walks you through a retail sales example.
If you're interested in this example and want to learn how it was used for time-series style modeling (using the LAGS function) watch this next video:
We encourage your feedback! Let us know how you are able to apply these methods to your own predictive modeling efforts.