The Best Laptop For Machine Learning And Data Science 2023 » EML (2024)

It’s no secret that data science and machine learning are two of the most in-demand skills right now.

Choosing the correct tool for the job is vital if you’re looking to make a career change into one of these fields or just starting out.

Now, as a disclaimer – we’re taking quite a bit of a different approach to our list than our competitors.

While I appreciate their research on this topic, I think they have yet actually to work in data science or machine learning.

The laptops you will see here will be all based on one premise, not just randomly researched laptops with good specs.

Don’t worry, I’ve got your back

Should I Use a Laptop for Machine Learning?

In the last 15 years, laptops have really blossomed into computation powerhouses.

I remember when I was a kid, you had your desktop at your house, and the laptop you used for school could do about 1/10 of your desktop computer.

Those days are long gone.

Laptops now are comparable in specs to desktop computers, and as battery life has improved in the last five years, they don’t need to be charged every 3 hours.

Now, the actual difference between a laptop and a desktop computer is the GPU.

While some laptops offer decent GPUs that can help speed up some of the heavier computations, those can be expensive and require custom fitting.

Thus, if you ever need to kick your machine learning analysis into overdrive and use powerful GPUs for complex models and large datasets, it’s much easier to connect your laptop (through the internet) to an online cloud provider likeGoogle Cloud,Amazon Web Services, orKaggle Workbooks.

With these services, you have all the power of a desktop with the mobility of a laptop – for a much lower cost.

You do not need a desktop computer in 2022 for machine learning. Suppose you ever run into a situation where the computation is too much for your laptop (which will sometimes happen). In that case, it makes way more economic sense to lease out the computation to a cloud provider (for pennies on the dollar) than spend thousands of dollars on a desktop setup.

What Are The Best Laptop For Machine Learning and Data Science?

Now that we’ve cleared up the misconception that you need the best GPU on the planet to start in Machine learning, there is one more thing we need to address.

If your goal when building your machine learning models is to put them into production, you will want a machine that closely mimics a production environment.

Why I Always Choose Apple’s OS (macOS) For Machine Learning

This is the best way to ensure consistency and usability when transferring your models and code to your server.

Since we know all production servers are Linux based (if you didn’t, you do now), this leaves us with two options.

1.) Linux

2.) macOS

The reason it’s these two options, and none other, is because they are Unix-Based operation systems.

Think of the Unix kernel as the tiny system that runs your code.

They will run the code the same way your production servers will.

Okay, back to the options – you could choose between Linux and macOS.

While Linux is an incredible operating system, it’s barebones and has a ginormous learning curve.

Instead of learning machine learning, you’ll spend most of your time learning Linux to dive into machine learning.

While this isn’t feasible, this leaves us with one last option – macOS.

Based on the Unix Kernel, Apple computers are incredible computers for writing code.

While I prefer Windows computers as a more cost-efficient machine, I’m always happy to pay the premium for Apple computers when writing code.

Our Two Choices For Best Laptop For Machine Learning and Data Science

1.) 2022 Apple MacBook Pro Laptop with M2 chip (13-in)

While any size between Macbook would work and get our vote, we prefer the 13in the model because it’s so easy to move around.

This Macbook Pro is a powerhouse, simply the ultimate laptop for data scientists, software engineers, and machine learning professionals.

It has the latest M2 processor (made by TSMC instead of Intel), 16GB of memory, and 1TB SSD.

The Retina display has True Tone technology, which adjusts the white balance to match your surroundings for more natural-looking images. (The screen is so crisp)

It also has Touch Bar ​​​​​​​and iTouch ID ​​​​​​​technology, which lets you use your fingerprint to unlock your computer or approve purchases on iTunes and the App Store. (Not something I use too much, but it is cool)

While this setup is awesome, you could get away with the 2021 M1 & a 512GB SSD if cost is a concern. (Don’t get lower than 16GB ram, especially if you want to do machine learning or use Chrome)

Pros:

• On-prem machine learning
• Optimal Production Ready Setup
• The mighty M2 processor with a 10-core GPU.
• Unix Based System
• MacBooks last 5+ years
• Insane Battery Life

Cons:

• MacBooks are overpriced for their specs
• M2 processor is only marginally better than M1 processor (Can get M1 to save $$)

2.) 2022 Apple MacBook Air Laptop with M2 chip (13.6in)

You guessed it, #2 on our list is the MacBook Pro’s younger sibling!

The 2022 Apple MacBook Air Laptop with an M2 chip is perfect for smaller computation tasks, working incredibly quickly to get jobs done.

The 512GB SSD Storage allows you to store all your datasets and models, while the backlit keyboard and 1080p FaceTime HD Camera make it easy to stay connected with friends, family, and teammates over slack.

With up to 18 hours of battery life, this laptop is perfect for on-the-go use.

Pros:

• Lightweight and Portable
• Easier to use on the go (airplanes, travel, etc.)
• Not a significant battery life dropoff over the MacBook Pro
• Cheaper than the MacBook Pro
• On-prem machine learning
• Unix based system
• Production-ready environment

Cons:

• Overpriced for its specs
• 8GB of RAM Hurts
• Since these laptops last 5+ years, spending more money and getting a MacBook Pro makes more sense.

Which Storage Type Is Best for Programming and Machine Learning?

RAM is the most important storage type for programming and machine learning (and probably computers in general).

When building your models, you will need to fit all your data into memory to be used – that’s your RAM.

Not only that but RAM should be prioritized during purchasing decisions.

While you won’t be able to upgrade most internal hardware on your laptops, it’s very cheap to offload SSD and hard drive data to an external cloud service for very cheap.

For example, Google offers 2TB of storage for…$10.00 a month.

This is not possible for RAM data; if it can’t fit into your computer’s on-prem memory, it can’t be used for modeling.

This is why RAM should be the main talking point regarding memory, not SSD’s size or storage options.

Is 8GB of RAM Enough for Data Science and Machine Learning, or do I need 16GB?

8GB is enough RAM for Data Science and Machine learning when you’re getting started, but eventually, you’ll want to move up to 16GB of RAM.

Before I upgraded to a MacBook Pro,I used this ASUSall the waythrough grad school.

While I was broke and had no options, I was able to get by with the 8GBs for grad school.

Actual picture of my wallet anytime I purchased anything during grad school.

Eventually, as my career progressed, I wanted to explore some more SOTA (state-of-the-art) models that aren’t feasible on 8GB, so I purchased a MacBook Pro.

Also, as a side note, Chrome is rough on 8GBs.

Which OS Is Better for Programming?

Linux is the best system for programming… if you know what you’re doing.

Since it’s incredibly lightweight and barebones, you’ll be able to run code at the speed of light.

You’ll have complete control over your computer, allowing you to mess with all system files and push your OS to its limits.

Since your production servers will have a UNIX-based system running on them, it’s a really consistent development environment.

However… With great power comes great responsibility.

Linux has an insane learning curve, and you can ruin your computer by running the wrong command in Linux.

What’s worse is even after spending TONS of hours learning Linux, installing it on a computer eliminates it from general usability.

This is why macOS is the best operating system for programming.

With macOS, you’ll be coding on a UNIX-based system (which will be similar to a production environment), have a strong and nimble enough OS that you have complete control over your dev environment, really no learning curve, and won’t crash your computer trying to upgrade permissions on a system file (speaking from experience).

Frequently Asked Questions

Are Apple laptops good for machine learning?

Apple Laptops are great for Machine Learning. While the M2 chip is a home run, the older intel processors were still really good.

If you’re going to get an older Apple laptop for getting into machine learning and want to save some money on your purchase, emphasize RAM (16GB) over the upgraded CPUs.

How good is the MacBook’s battery life during modeling?

The MacBook’s battery life is excellent for data science and machine learning. The 2022 M2 MacBook Pro has up to 20 hours of battery life, which is plenty of time for you to run a model or two and find your charging cord. MacBooks currently dominate the battery life category due to the leanness of their operating system over their competitors.

Are MacBooks Good Enough For Deep Learning?

MacBooks are great for Deep Learning. Again, the barrier to machine learning is RAM size, not your CPU. And, as we went over earlier, it’s VERY cheap to lease out intense training to an online GPU system like Kaggle or Google. So, if you follow our instructions above and scoop up a 16GB system, you’ll be on your way to becoming a deep learning pro.

How to Find Cheap Laptops for Machine Learning

Sometimes cheap is more expensive in the long run. Remember, your time is not infinite. While you can purchase a cheap laptop and get quickly get started with machine learning, you may end up paying more in time fixing things or waiting for models to run and finish. If cost is a serious concern, consider a refurbished MacBook Pro with 16 GB RAM. You can check some outhere.

Should I Get A Desktop Instead Of A Laptop for Machine Learning?

The days of Desktop dominance are long gone. If you’re interested in gaming, you should purchase a Desktop – since you can load it up with a super expensive GPU.

If you don’t play games, I do not see the point of purchasing a Desktop computer. Purchase one of the laptops recommended above – and if it’s ever not powerful enough for some machine learning or data science task, use a cloud service (for pennies on the dollar) to finish it up.

You can also just check our guide if you’re just in the market for a keyboard for data science.