# Regularization or Add more data

**URL:** <https://community.deeplearning.ai/t/regularization-or-add-more-data/205747>\
**Category:** Advanced Learning Algorithms\
**Tags:** week-module-3\
**Created:** [September 17, 2022, 1:14pm UTC](https://community.deeplearning.ai/t/regularization-or-add-more-data/205747 "2022-09-17T13:14:39Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Med-akraou](https://avatars.discourse-cdn.com/v4/letter/m/a9adbd/32.png) [@Med-akraou](https://community.deeplearning.ai/u/Med-akraou)\
**Post date:** [September 17, 2022, 1:14pm UTC](https://community.deeplearning.ai/t/regularization-or-add-more-data/205747/1 "2022-09-17T13:14:39Z")

</div>

When we have overfitting or variance what should we start with to improve the model? add more data or regularization ?

---

<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:** [September 17, 2022, 3:35pm UTC](https://community.deeplearning.ai/t/regularization-or-add-more-data/205747/2 "2022-09-17T15:35:54Z")

</div>

Hello @Med-akraou,

If we already have more data, I believe we will add them sooner or later, and the rest of the hyperparameters can change because of the additional data, so I would suggest to add them first if they are already available, however, if we will need time to collect more data, then we probably should go for regularization first.

Raymond

---

<div class="post-metadata">

**Author:** ![Med-akraou](https://avatars.discourse-cdn.com/v4/letter/m/a9adbd/32.png) [@Med-akraou](https://community.deeplearning.ai/u/Med-akraou)\
**Post date:** [September 17, 2022, 3:57pm UTC](https://community.deeplearning.ai/t/regularization-or-add-more-data/205747/3 "2022-09-17T15:57:05Z")

</div>

Makes sense. Thank you very much @rmwkwok