View profile

I'm back

I'm back
By Santiago • Issue #20 • View online
My vacation was great.
Good food, tall buildings, and a lot of walking around. New York never disappoints!
Now I’m back to the quietness of the suburbs. And to writing and thinking about machine learning stuff.
I definitely missed it!

I took this picture from the Empire State Building. With my iPhone.
I took this picture from the Empire State Building. With my iPhone.
The accuracy of a regression model
How do you compute the accuracy of a regression model?
I remember asking this question myself when I started. I’ve also heard of interviewers trying to trick people with it.
Here is an explanation of how to think about this.
How do you compute the accuracy of a regression model?

Somebody asked this question.

It reminded me of how much I struggled to understand the difference between two of the most popular machine learning tasks.

What's happening here?
Speaking about evaluating regression models...
Here is a comparison between RMSE and MAE.
I’ve used both methods to evaluate models. They are slightly different, but understanding why is really important.
(This is one of those things that, when used correctly, blows everyone else’s mind because it’s really clever.)
Is 10 twice as worse as 5? Sometimes it is, but sometimes it's even worse.

This is the question I always ask myself when deciding how to penalize my models.

Read on for more details and a couple of examples:

↓ 1/11
How to choose a good machine learning model?
Last week I shared a Medium article that I wrote about choosing a good machine learning model.
Some of you mentioned Medium’s paywall. I know this is a problem.
Here is the same content on Twitter. The good news is that you get the goodies in a much more straightforward manner!
A good machine learning model is not only about performance.

In Kaggle competitions, that might be all you need. In real-life situations, it's just a factor to keep in mind.

Thread: How to choose a good machine learning model?
Links and articles
You noticed that I’ve been sharing a lot of links to separate blog posts or Twitter threads.
That’s not how I started this newsletter: it was always a full article.
There was a problem with that format: it forced me to write something specific for the newsletter. It wasn’t about sharing something good but about sharing something new.
This sucked, so I decided to change it.
I’ll keep sharing the best content I write regardless of where it lives. That’s the right thing to do.
If you have an idea that you’d like me to explore, reply to this email and let me know.
See ya next week!
Did you enjoy this issue?

I'll send you an email whenever I publish something important. I promise you don't want to miss this.

In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue