Feature Scaling Part 2: issue with feature scaling to [-1;1]

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
1

Greetings!

During the lecture Andrew states:
As a rule of thumb, when performing feature scaling, you might want to aim for getting the features to range from maybe anywhere around negative one to somewhere around plus one for each feature x. **But these values, negative one and plus one can be a little bit loose**. If the features range from negative three to plus three or negative 0.3 to plus 0.3, all of these are completely okay.

What is the issue with feature scaling to [-1;1]?
Why this range is considered as a loose one?

Is feature scaling to [0;1] a better option?

Thank you.

2

When you reference a lecture, it’s helpful if you give both the lecture title and the time mark.

Andrew lectures in a very intuitive way, rather than assuming the audience has a high degree of math skills.

In this case, what he means by “these values can be a little bit loose” is that -1 and +1 aren’t hard limits. You can use any range of small negative values that are symmetric around zero.

3

Sure!
Actually, the lecture title is in subject title.
The time mark for this part is 4:09.

4

what he means by “these values can be a little bit loose” is that -1 and +1 aren’t hard limits

Got it.
I initially interpreted “loose” as “not suitable” or “not appropriate”.

5

I realize that. It’s why my response said to provide both the title and time mark.

1 Like
6

I will surely do this next time.

1 Like
7

Thanks. It just makes the mentor’s tasks a little easier.