At times you will want to run models on a smaller dataset than how the project was trained. For example, if your dataset has over 1000 features, you can create your project with just a subset of features and then add new features as you go.
In this tip, I show how you can create a project with one feature list, and then run it on another feature list, all through the DataRobot UI. (You can find information for feature lists in the documentation.)
Then, from the Feature List dropdown you can select that new feature list and view its content.
Now, when you’re ready to start the project, just choose that new feature list and press Start. DataRobot will use your new feature list to build models.
I hope this tip was helpful! I have a couple more tips in the works and will post them soon. If you have tips to share with others, please post them!
- @dalilaB
can you share another tip for this but for using the API? great tip thank you
@FranksT pls do take a look at the DataRobot University API Python and R classes listed below where we cover feature selections as well as other topics:
https://university.datarobot.com/advanced-datarobot-with-python
https://university.datarobot.com/api-i-r
We will soon publish one using R or Python
As promised, here is the link on using API