You finished building your model, and you already deployed it.

You ain't done yet, unfortunately. You are arguably only half the way through.

The performance of machine learning models degrades over time.

Let's talk about this.

A quick introduction to a machine learning model:

X → y

• X: Input data
• y: Output prediction
• →: The relationship that the model learned between X and y.

Read 11 tweets

Santiago

@svpino

24 May

Today, Python 🐍is the best language you can learn if you want to get into machine learning.

It has many things going for it, but here are 2 very important:

• It's very flexible.
• It's very popular.

This, however, can change.

I get a lot of questions about Python versus R.

R is great, but it doesn't have the flexibility that Python does.

Learn Python, and you can use it for more than machine learning: you can build your backend, MLOps, DevOps, tooling, etc.

One language. Multiple uses.

If you look at the literature today, it's Python all the way down.

If you look at libraries and frameworks, they are Python-first with a few spinoffs on different languages.

It's hard to go against the current.

Read 6 tweets

Santiago

@svpino

23 May

The best advice I've ever gotten:

"Nobody remembers who came in second place."

I played a lot of sports. I was taught that winning is the only thing that matters. Participation trophies weren't a thing back then.

That's how I'm wired.

↓

Many people prefer to approach their careers with a different, less competitive mentality.

I respect that. I just prefer a different approach.

Complacency is something that terrifies me, and I've found that healthy competition keeps me away from it.

↓

"Never compare yourself to others." ← This may be good advice, but it doesn't work for me.

I've tried to use yesterday's version of myself to try and improve every day. But this is not enough, too little, and not fun.

↓

Read 4 tweets

Santiago

@svpino

22 May

I'll blow your mind with a technique you aren't using yet.

Sometimes, you want your system to do exactly the opposite of what your machine learning model thinks you should do.

Let me convince you. ↓

I'm going to start with a nice problem:

Imagine a model that looks at a picture of an electrical transformer and predicts whether it's about to break or not.

Don't worry about how the model does this. We are going to focus on the results instead.

There are 4 possible results for this model:

1. It predicts a bad unit as bad.
2. It predicts a bad unit as good.
3. It predicts a good unit as bad.
4. It predicts a good unit as good.

#2 and #3 are the mistakes the model makes.

Read 21 tweets

Santiago

@svpino

20 May

Machine learning is not what you think.

Here are five levels of automation:

• L0. Human-only
• L1. Shadow mode
• L2. AI-Assisted
• L3. Partial automation
• L4. Full automation

Everyone thinks L4, but that's not necessarily the end game.

Let's break these down. ↓

• L0. Human-only

Obviously, there's no automation at this level. Human makes all the decisions, so this is a manual process.

This is usually the initial state of any system right before we start automating it.

• L1. Shadow mode

As we introduce AI, a useful step is to deploy it parallel to the existing manual process.

We route requests to both the person and the system and get answers from both.

At this level, the AI system is not involved in any decisions.

Read 9 tweets

Santiago

@svpino

19 May

Wanna get into Machine Learning but don't know where to start?

Join the other 1,400 people that have watched my course. It's gotten well over 100 5-star reviews!

$5 for the next 24 hours only!

Your full money back if you don't like it.

gum.co/kBjbC/five

If you want to support my content but don't have the money or don't need the course, a retweet/like of the original tweet goes a long way!

If you can't afford the course, comment under the original tweet, and I'll send you a link to a free copy.

Thanks for the support!

18 more hours until the price goes back to normal.

Read 5 tweets

Share this page!

Santiago

Try unrolling a thread yourself!

More from @svpino

Santiago

Santiago

Santiago

Santiago

Santiago

Santiago

Did Thread Reader help you today?

Like this author's thread?