I have 2 models with different dependent variables. The independent variables are more or less the same. Model 1: performs well, but I would like to find feature importance/ significancy out from the model constructed. Model 2: overfits, I need to solve/ reduce over fitting.
What would the best feature selection method (filter, wrapper or embedded methods) for models 1 and 2?
Dear Shaima Mohammed Alghamdi
You will find this article helpful in finding your queries:
https://www.analyticsvidhya.com/blog/2016/12/introduction-to-feature-selection-methods-with-an-example-or-how-to-select-the-right-variables/#:~:text=Filter%20methods%20measure%20the%20relevance,training%20a%20model%20on%20it.&text=Filter%20methods%20use%20statistical%20methods,wrapper%20methods%20use%20cross%20validation.
Hello Shaima Mohammed Alghamdi ,
If I understand correctly, your two model differ in their respective hyperparameters. These include resolution of the input, depth and size of the layers.
This paper is very interesting in that case: Preprint EfficientNet: Rethinking Model Scaling for Convolutional Neu...
EfficientNets have optimal hyperparameters and could solve your overfitting problem.
Cheers, Raoul
Have a look here for some tips
https://medium.com/greyatom/what-is-underfitting-and-overfitting-in-machine-learning-and-how-to-deal-with-it-6803a989c76
https://medium.com/analytics-vidhya/feature-selection-techniques-2614b3b7efcd
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering