# Feature Scaling and learning rate lab

**URL:** <https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455>\
**Category:** Supervised ML: Regression and Classification\
**Tags:** week-module-2\
**Created:** [August 23, 2023, 12:00am UTC](https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455 "2023-08-23T00:00:19Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Abdelrahman\_Osama2](https://avatars.discourse-cdn.com/v4/letter/a/c5a1d2/32.png) [@Abdelrahman\_Osama2](https://community.deeplearning.ai/u/Abdelrahman_Osama2)\
**Post date:** [August 23, 2023, 12:00am UTC](https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455/1 "2023-08-23T00:00:19Z")

</div>

Here we train the model using normalized data, so we get normalized parameters.  
when making predictions I think the result will also be normalized, so how to make the nomalized predicted value real? Also, is the predicted price here normalized or real?

 ![2023-08-23_02-57](https://global.discourse-cdn.com/dlai/original/3X/d/7/d7b1829d14a30fd20ed74b070862bca687d2f35a.png)

---

<div class="post-metadata">

**Author:** ![TMosh](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/tmosh/32/9321_2.png) [@TMosh](https://community.deeplearning.ai/u/TMosh)\
**Post date:** [August 23, 2023, 12:30am UTC](https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455/2 "2023-08-23T00:30:41Z")

</div>

To get a prediction, you apply the same normalization (mu and sigma) to the features, then compute the value.

It’s the real result - there are no normalized outputs.

---

<div class="post-metadata">

**Author:** ![rmwkwok](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/rmwkwok/32/126977_2.png) [@rmwkwok](https://community.deeplearning.ai/u/rmwkwok)\
**Post date:** [August 23, 2023, 12:31am UTC](https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455/3 "2023-08-23T00:31:46Z")

</div>

Hello @Abdelrahman_Osama2,

Welcome to our community!

Before training, we normalize the features, and generally we do not normalize the labels.

After training, like you said, the weights are adapted to the normalized features, but since the model is trained to model those labels that have never been normalized, the predictions are in their original scale - we do not need to do anything to unnormalize them.

Here, if the labels are in thousand dollars, then the predictions are in thousand dollars. If the labels are in million dollars, then the predictions are in million dollars.

Cheers,  
Raymond

---

<div class="post-metadata">

**Author:** ![Abdelrahman\_Osama2](https://avatars.discourse-cdn.com/v4/letter/a/c5a1d2/32.png) [@Abdelrahman\_Osama2](https://community.deeplearning.ai/u/Abdelrahman_Osama2)\
**Post date:** [August 23, 2023, 10:07pm UTC](https://community.deeplearning.ai/t/feature-scaling-and-learning-rate-lab/425455/4 "2023-08-23T22:07:53Z")

</div>

Thanks so much. I finally got it.