Optional Lab: Feature scaling and learning rate. Z-score normalization

KiraDiShira · September 17, 2022, 1:07pm

From what I understand, the range of #Bedrooms, #Floors and maybe Age looks acceptable. While size is different order of magnitude greater so it needs to be rescaled.

But in the lab all features are rescaled.

Why?

Maybe if I want to scale a feature I need to scale all?

Thank you all

rmwkwok · September 17, 2022, 3:33pm

Hello @KiraDiShira,

We don’t have to scale all when we want to scale one feature, but i would always suggest to scale all since the key is to make sure all features having similar scales and scaling all features does exactly that.

Raymond

Topic		Replies	Views
Question about rescaling Supervised ML: Regression and Classification week-2	4	503	July 7, 2022
Do we scale the features assuming all the features have the same amount of influence on the result? Supervised ML: Regression and Classification week-2	5	409	August 17, 2023
Feature scaling: scaling one mean scaling all? Supervised ML: Regression and Classification week-2	1	427	June 5, 2023
Feature Scaling - When to Scale Supervised ML: Regression and Classification week-2	3	494	July 13, 2023
Is my understanding of Feature Scaling correct? Supervised ML: Regression and Classification week-2	3	526	August 12, 2022

Optional Lab: Feature scaling and learning rate. Z-score normalization

Related topics