Santiago Profile picture
23 Feb, 7 tweets, 2 min read
When you start with machine learning, it's tempting to learn as many different algorithms and methods as possible.

This is not the best approach. This will not make you the best you can be.

[1 / 5] 🧵👇
Instead, focus on understanding the power of representations and getting as good as you can at feature engineering.

Feed garbage to your fancy algorithms and they will give you garbage back. No exceptions.

[2 / 5]
"Representation" is the process of mapping data into useful features.

"Feature engineering" is the process of determining which features might be useful in training a model.

There's a lot of creativity involved here! The time you spend will pay you back in spades.

[3 / 5]
Most of the time, if you put in the work to select (and create) the best possible set of features, the algorithm you end up using becomes secondary.

I like to call this a "data-first approach." Not sexy, but extremely powerful.

[4 / 5]
At this game, the best is not whoever knows the most, but whoever can think creatively.

And if you are looking for some help, follow me, and every week I'll help you navigate this thing from doing it, failing at it, learning, and fixing it.

[5 / 5]

I put this thread about Feature Engineering some time ago:

Excellent point: breath is needed to bypass interviews.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Santiago

Santiago Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @svpino

25 Feb
Today, let's talk about two key data transformations we constantly use in machine learning:

▫️ Label encoding
▫️ One-hot-encoding

But let's not just talk about them, but try to build some intuition about why they are important.

Grab a coffee, and let's start! ☕️🧵👇
Imagine we have a dataset with two features:

▫️ "temperature" — a numeric value.
▫️ "weather" — a string value.

You should feel uncomfortable with this dataset right off the bat: machine learning algorithms usually don't like to work with non-numerical data.

[2 / 15] Image
To set the record straight, some algorithms don't mind non-numerical data.

For example, certain Decision Tree implementations will be fine with the "weather" feature from our example.

But a lot of them can only work with numbers.

[3 / 15]
Read 15 tweets
24 Feb
Let's do a line-by-line analysis of this deep learning model and truly understand what's going on.

This model identifies handwritten digits. It's one of the classic examples of machine learning applied to computer vision.

🧵👇 Image
First of all, we load the MNIST dataset. This dataset contains 28x28 images showing handwritten digits.

This dataset is so popular that Keras built a utility to load it with a single line of code.

The function returns the dataset split into train and test sets.

[2 / 24] Image
x_train and x_test represent the train and test sets containing the features: the 28x28 matrix representing the image.

If we print both sets' shapes, we will get 60,000 train images and 10,000 test images.

[3 / 24] Image
Read 24 tweets
22 Feb
How do you know you're ready to apply for a data science or machine learning job?

▫️ Remove from your resume everything related to your education: schools, tutorials, certificates, etc.

After getting rid of all of that, would somebody hire you by looking at what's left?

If the answer is no, then you aren't ready.

Your primary asset is the experience you bring to the table. If you have none, finding a job will be hard.

I'm not talking about "years" of experience but your ability to find solutions to problems.

Experience is not necessarily related to having a job either.

In fact, a job may become detrimental to your experience because you'll have to work on something specific for too long.

Just focus on solving problems on your own. Then talk about them.

Read 7 tweets
21 Feb
Go to college. Send your kids. Celebrate those that make it happen.

College is a good thing. If you can afford it, do it.

Most people telling you that college sucks went to college. Most people that didn't go whish their kids would.

You won't replace college with YouTube videos, or reading books, or following tutorials.

Some people may. Most people won't.

Yes, the knowledge is all out there, but college is just not about learning new things.

College doesn't guarantee you a job but look at the statistics of median income and unemployment among those that went and those that didn't.

The numbers should tell us something.

Read 8 tweets
20 Feb
25 popular libraries and frameworks for building machine and deep learning applications.


▫️ Data analysis and processing
▫️ Visualizations
▫️ Computer Vision
▫️ Natural Language Processing
▫️ Reinforcement Learning
▫️ Optimization

A mega-thread.

🐍 🧵👇
(1 / 25) TensorFlow

TensorFlow is an end-to-end platform for machine learning. It has a comprehensive, flexible ecosystem of tools and libraries to build and deploy machine learning-powered applications.
(2 / 25) Keras

Keras is a highly-productive deep learning interface running on top of TensorFlow. It provides essential abstractions and building blocks for developing and shipping machine learning solutions with high iteration velocity.
Read 20 tweets
19 Feb
I'm sad to watch many developers working 80-hour weeks to get one inch ahead of everyone else.

And yet, they are missing the biggest opportunity of their lives right under their noses.

Today, you don't leap ahead by learning another framework, watching another tutorial, or building another web page.

That's incremental improvement. Important, but not extraordinary.

Hours don't mean anything, and everything you add to your portfolio will be obsolete in the next couple of years.

What's really going to move the needle is the impact of your work. It's how you change and influence those around you.

Read 10 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!