C2W2 - Scaling - "the model learns the right weights"?

Course Q&A Machine Learning Engineering for Production(MLOps) Machine Learning Data Lifecycle in Production
1

Can anyone explain more clearly what it means when saying that benefit of scaling for each feature is that the model learns the right weights?

2

Hi @Dinh_Tr_n ,

Thanks for your question.

We scale the features to prevent certain features (which operate in ranges with high values) dominating the prediction. Scaling ensures that each feature approximately carries equal weight in the model. There is plenty of material online going deeper into feature scaling , for instance:

Hope this helps answering your question…

Best Regards,
Maarten

1 Like