How do I make a model select more (or less) features / variables?


From the augmentation screen, you can configure the model's Dimensionality Reduction (Augmentations >Edit > One AI Configuration > Dimensionality Reduction > Override). Dimensionality reduction refers to the process of reducing the number of input variables or features in a dataset while preserving the most relevant information. However, you can also increase the number of features here. 

Here, you can change your filter method and input the num features, which is basically the maximum number of variables your augmentation can select. You can also change your wrapper methods here and input your min features, which is the minimum number of variables your augmentation can select. For example, if you want the model to select 10-15 features, you would configure it like the image below. Just make sure that your wrapper methods min features is less than or equal to the filter method num features, or your model will error. 

Dimensionality Reduction Configuration

Was this article helpful?

0 out of 0 found this helpful



Please sign in to leave a comment.