How Outliers Hijack Your Machine Learning Model

Added: 2026-05-27

[00:00:00]Now, here's where things get interesting. Let's add your lucky day data. Say the spots you hit were around here. And suddenly, we add seven legal-sized crabs.

[00:00:14]Looks like you moved further out because you weren't catching anything big closer to shore.

[00:00:19]Now, we recompute MSE.

[00:00:23]A and we follow the curve all the way down.

[00:00:31]Perfect. Now, the line shifts like this.

[00:00:34]If you just look at this result, the model gets completely fooled. It looks at that lucky data and starts acting [music] like, "Whoa, wait. Location actually has a strong positive effect on crab size."

[00:00:48]See? Look at the weights. One literally blew up. It's more than five times what usually is. But, based on your experience, those seven crabs, total fluke. Not normal at all. We [music] definitely don't want those few points hijacking the whole model.

[00:01:04]So, how do we reel it back in?

[00:01:07]Was this helpful? There's even more exciting content waiting for you in our full lesson. Click the link below now, and let's take the next step in our growth together.