The Best Computer for Data Science Beginners

The Best Computer for Data Science Beginners

In this video, I give you my thoughts on the best computer for data science beginners. I talk about the hardware components of a computer and how those relate to your data science performance. At the end, I give my thoughts on the best computer for you to purchase.

The cpu is in charge of the processing power of the computer. It mostly controls how fast processing for normal data science operations is.

The ram stands for random access memory. It determines how much data can be processed in a single batch.

You will also need a hard drive. I recommend stead state memory. This doesn’t have much impact on performance of your models, so you don’t need to go crazy here. This is the cheapest type of memory though so you can scale up at a low cost.

The last main component that is relevant for data science is the GPU. NVIDIA recently sent me a Titan RTX, and I am excited to be experimenting with that in future videos. For deep learning, you should really have at least 6 gb of gpu memory. If you are playing graphics intensive games or editing videos, I recommend increasing this further.

My final recommendation: Your computer really doesn’t matter much. Data science is done mostly on the cloud making your computer relatively irrelevant. You have access to top of the line GPU’s and even TPU’s on or through google colab. In my opinion, the best computer for data science is the one that you can afford.

If you still want to do something locally, I recommend the following requirements

CPU (4 cores): Ryzen 5, 7, 9 should suffice
Ram: Minimum 8gb but 16 is preferred
HD: 256gb
GPU: NVIDIA with 6+ GB of memory (4 + gb if if laptop)

My laptop:

Budget PC Parts
CPU: Ryzen 5 ($150)
RAM: Corsair 16gb ($75)
HD: WD SSD 480 GB ($55)
GPU: GTX 1660 ($230)
Total: $510 (before case, motherboard & power supply ~$200)

Mid Tier PC Parts
CPU: Ryzen 7 ($284)
RAM: Corsair 32gb ($150)
HD: WD SSD 480 GB ($55)
GPU: RTX 2070 ($640)
Total: $1129 (before case, motherboard & power supply ~$200)

High Tier PC Parts
CPU: Ryzen 9 ($423)
RAM: Corsair 32gb ($150)
GPU: Titan RTX ($2500) (borderline overkill)
Total: $3252 (before case, motherboard & power supply ~$200)

If you do decide to build a pc, I recommend using to make sure your parts are all compatible!

#DataScience #KenJee

โญ• Subscribe:
๐ŸŽ™ Listen to My Podcast:
๐Ÿ•ธ Check out My Website –
โœ๏ธSign up for My Newsletter –
๐Ÿ“š Books and Products I use – (affiliate link)

Partners & Affiliates
๐ŸŒŸ 365 Data Science – Courses ( 57% Annual Discount):
๐ŸŒŸ Interview Query –

๐ŸคMy Twitter –
๐Ÿ‘” LinkedIn –
๐Ÿ“ˆ Kaggle –
๐Ÿ“‘ Medium Articles –
๐Ÿ’ป Github –
๐Ÿ€ My Sports Blog -

Check These Videos Out Next!
My Leaderboard Project:
66 Days of Data:
How I Would Learn Data Science in 2021:

My Playlists
Data Science Beginners:
Project From Scratch:
Kaggle Projects:


  1. Piyush Thakur on May 27, 2021 at 4:40 am

    I got Acer Aspire 7 and is perfectly doing its job .

  2. Thahir Hanif on May 27, 2021 at 4:40 am

    Thanks for making this video Man …!! I have never own a Laptop or system before it’s more helpful to understand the requirements of a system….! Actually I’m also planning to buy a Laptop for learning DS big thumbs up to u man…..! But the problem is there is no such laptops under the price range of 500 $ with the specs u would mention in the video ๐Ÿ˜”๐Ÿ˜” could u pls suggest some if u don’t bother abt the mentioning……!

  3. Ken Jee on May 27, 2021 at 4:40 am

    Thanks for watching everyone! I hope this video was helpful to you. If you’re new to data science, I recommend watching this new playlist I just put together: . I took all of the best videos I made for beginners and threw them in here!
    If you just want to jump into my recommendation you can skip to 2:50!

  4. Joseff Tan on May 27, 2021 at 4:42 am

    Please upload more data science project as well like krish naik ๐Ÿ˜๐Ÿ˜๐Ÿ˜

  5. Miguel Rosales on May 27, 2021 at 4:43 am

    Hey, Ken. I’m using Asus Vivobook. What can you say about it? Thanks

  6. TaylorKong on May 27, 2021 at 4:44 am

    were you a rower? heard you say "steady-state" drive and threw me for a loop lol

  7. Mashoor Araf on May 27, 2021 at 4:44 am

    Hey ken. I know it’s a off topic for this video but I only know you an IT expert who can answer my questions. I want to learn android development . I am already a web developer. But I am not sure if I need a computer science degree to become a android developer. So can you tell me do I need a computer science degree in order to become a android developer or can become a self-taught android developer.

  8. Elbow Slam on May 27, 2021 at 4:48 am

    Nice home exterior view u have.

  9. Denis Belikov on May 27, 2021 at 4:50 am

    Ken Jee, tell me please! SSD on SATA 3.0 for what kind of work in data science we can use? And for SSD on NVMe the same question) hope you understand what do i mean) Thanks a lot for your videos, interesting to listening to you!

  10. Pranav Arora on May 27, 2021 at 4:52 am

    Awesome video ken, just a quick note there , it would have been HDD and not HHD :p. Else it was yet again a great watch

  11. moot talk on May 27, 2021 at 4:54 am

    I am sorry, 256G of hard drive for data science ? I really think you should do a few data science projects before you start recommending these specs. 256G even for a regular system might not be enough (even considering the fact that lot of data is stored in cloud). For data science you should be looking at TBs of hard drive space.

  12. Gregory Horne on May 27, 2021 at 4:57 am

    I recently switched from a Hewlett Packard Spectre 13 (8 GB RAM, 1 TB M.2 2880 SSD, Intel i5) to a Lenovo ThinkCentre M700 (16 GB RAM, 512 GB SATA III SSD, Intel i5-6400T) for data science. The impetus for the migration was my need for an ergonomic keyboard because modern notebook computer keyboards are terrible for extended typing sessions. Eventually, I will add an NFS server providing easily expandable file storage but that is likely a project for next year as part of a larger re-architecture project for my home office network and computing hardware.

  13. Matias Pappalardo on May 27, 2021 at 4:58 am

    Thanks for sharing. Do you think is better a desktop or a lapatop?

  14. ์ง„ํฌ๊ณค on May 27, 2021 at 4:58 am

    Can you show us how to get/use GPU resources from online such as Amazon service or Google platform?
    I tried, but felt a little hard

  15. Fahad Reda on May 27, 2021 at 5:01 am

    a year ago i built a computer for ML and DL , Thanks Ken , another great video as usual

  16. Kleos on May 27, 2021 at 5:03 am

    This video came at the perfect time! Iโ€™m leaving for grad school soon doing an MS in Data Science and Analytics, and my laptop is about 10 years old and hella slow. I desperately need a new one but as a student I donโ€™t have the budget for another Apple product.

  17. Dereck de Mรฉzquita on May 27, 2021 at 5:05 am

    Thanks for the content Ken but SSD stands for solid state drive not stead(y).

  18. Keshav Balachandar on May 27, 2021 at 5:06 am

    Hey Ken, how’re you? Great video once again ๐Ÿ™‚
    Apart from MacBook and Dell XPS what are the other laptop suggestions you have?

  19. jacob richard on May 27, 2021 at 5:07 am

    Hey Ken,
    How do you go about classifying data as unknown aka outside the distribution that the training classes fall into? My neural net softmax probs are high for data that should be unknown

  20. Andrey Protas on May 27, 2021 at 5:09 am

    I found this video to be extremely basic and vague (and the video doesn’t quite match the contents: you show very expensive components and then speak how it’s not really necessary to have powerful PC for beginners). The general recommendations are fine, but there is no explanation as for when and why you would want to get anything better than Ryzen 5.

    What tasks are particularly CPU focused, and will benefit noticeably from e.g. Ryzen 9 processor? If it can save you several hours of computing a day, it can be a big deal, but if it says less than a hour, it’s mostly pointless.

    The same goes for graphic card. Nvidia RTX Titan seems extremely overpriced for what it does. From what I can tell, it’s only slightly better than RTX 2080 Ti, and cots twice as much. Is there even any reason to get it? Maybe some additional optimization for Data Science tasks? So far it looks extremely cost ineffective, even for highest end PC. Might as well get something cheaper for now and then upgrade to RTX 3000 when it launches.

    Again, unless it can save several hours of computing every day, I don’t see how it makes any sense at all.

    Additionally, I wonder how impactful RAM memory speed is for Data Science tasks. Does it make sense to overclock your RAM, or is it just a waste of time and effort?

    Finally, I’m not sure if it’s correct to simply ignore Motherboard and PSU (You listed a 500$ PC, and then you mentioned 200$ for motherboard and other components, which practically increases the price by 50%). Again, depending on how important storage speed is, it might make sense to get a more expensive motherboard, such as B550 or X570. However, if it’s not as important, and regular NVME SSD is more than good enough, then good B450 board is plenty powerful. As for storage Samsung NVME SSDs are way overpriced, so I wouldn’t recommend one personally. Good ADATA NVME costs just 70-75$ for 500 GB, and is only marginally slower.

    My recommended budget build based in my understanding will depend on the country, but in short here doesn’t seem to be any significant differences from general gaming build recommendations.

    For processor I’d get Ryzen 5 1600 AF if it’s available for under 110$, otherwise Ryzen 5 3600 is the best choice.

    For memory if speed is not super important for Data Science tasks I’d recommend Crucial Ballistix Black 2×8 GB 3200 MHz. Even if it is this kit has good overclocking capabilities. Wouldn’t get anything below 16 GB, as Chrome is a memory hog, and you likely need a lot of tabs open for data science projects.

    For motherboard I’d recommend B450 Mortar Max or Tomahawk Max, as those support upgradability. For cheaper options B450 PRO-VDH Max is also good.

    For SSD ADATA SX8200 Pro 500/1000 GB is the best choice (depending on your preferences regarding regularly cleaning your PC). I wouldn’t get anything below 500 GB myself, since this will likely cause major inconveniences.

    For PSU anything 550-650W is good enough. I wouldn’t get anything below Gold certification if your PC runs for 10+ hours every day. If available at reasonable price, Deepcool DQ-650M is a good choice.

    Monitor doesn’t really matter for Data Science, so anything cheap and sturdy will do.

    Graphic card is a tricky one. I’d probably try to look for a good used option for a budget build, but it’s hard to find a reliable seller. I assume Nvidia cards are easier to use for Data Science, as AMD cars get less support. Used GTX 1060 6GB is a bare minimum I’d recommend, but if you can afford a more expensive one, 2070 Super looks like an optimal choice in terms of cost to performance. Personally I’d get 1060, and then upgrade to RTX 3070 when available.

    As for the case, anything above 60$ should be fine. Got myself new MSI MAG case, it’s quite, runs pretty old and comes with fans already pre-installed. More premium options are Phantex P400A, and Meshify C.

    Secondary components like motherboard, PSU and case may seem unimportant, but they ultimately affect performance, reliability, upgradability and longevity of the system as we.

    Still would like to know how important memory speed is for Data science tasks. 16 GB 2400 MHz and 16 GB 3800 MHz are 2 completely different kits for some tasks.

  21. chai-lattae on May 27, 2021 at 5:10 am

    Great content as usual, thank you for your insight!

  22. Mario on May 27, 2021 at 5:10 am

    That GPU looks like a monster :O
    Can’t imagine the performance difference with that!

  23. Piyush Zope on May 27, 2021 at 5:10 am

    @Ken Jee I have one question on this… Even if someone have good knowledge and some experience but companies give different role instead of what we want so what to do in that case?

  24. gp1497 on May 27, 2021 at 5:11 am

    Great video! Always informative.

  25. MTanmay on May 27, 2021 at 5:11 am

    A quick question
    Is it possible to run CUDA on an AMD CPU ? (Yes *CPU* , not talking about GPU)

  26. JT Vader on May 27, 2021 at 5:12 am

    Thank you so much Ken for your awesome videos

  27. IMPOSSIBLE THINGS on May 27, 2021 at 5:14 am

    Nice guidance Video.
    Video quality is increased so greatly.

  28. ANUMEHA SHRIVASTAVA on May 27, 2021 at 5:15 am

    Thank you ken ๐Ÿ™‚

  29. Tirupati Panaganti on May 27, 2021 at 5:17 am

    Great video ken . Keep them coming.

  30. Barneyy on May 27, 2021 at 5:19 am

    I bought a $699 Lenovo IdeaPad flex 5 laptop as i was a newb and didn’t know about the gpu requirement. This one has an AMD Radeon gpu. I have the option to cancel the order. Should i cancel it?

  31. Muhammad Hashir on May 27, 2021 at 5:20 am

    Sir, I request you to interview Joma Tech

  32. Avicii on May 27, 2021 at 5:20 am

    Thanks ken..!

  33. Javi Garcรญa-Ripoll on May 27, 2021 at 5:21 am

    256GB is NOT enough! If you had to install several operating systems you’d be screwed!

  34. RodStremel on May 27, 2021 at 5:21 am

    Ken, what about the keyboard? I see you’re using a Durgod Hades 68. Do you think a mechanical keyobard is worth the investment? Also, do you think that a TKL keyboard work just fine for doing data science?

  35. behzad h on May 27, 2021 at 5:22 am

    Hi Ken. It will be great if you share your opinion about the new M1 MacBooks for the field of data science, especially for beginners. Thanks!

  36. Reid on May 27, 2021 at 5:27 am

    AMD Ryzen series is awesome, I just rebuilt with that and an AMD Radeon RX 5700 XT. Absolutely awesome.

  37. Ilya Ginsburg on May 27, 2021 at 5:28 am

    I’ve got a Ryzen 7 computer with 32G RAM and NVIDIA 2080 Ti, since I’m going to almost exclusively work with neural networks. My professor has 128G RAM and claims that isn’t enough for the real-life tasks ๐Ÿ™‚
    I don’t really like Colab for the serious work. First, the session expires pretty quickly, and you should fall to the shameless tricks to keep it. Second, transferring gigabytes of data for training via Internet may nullify all the advantages of TPUs. Colab is great for the small learning or experimental projects, but it can’t train a BERT-based network that required more than a day on my 2080 Ti.

  38. Spade on May 27, 2021 at 5:28 am

    can i use a rx 580

  39. Israel Bayode on May 27, 2021 at 5:28 am

    Thanks, Ken, for this! You’re simply amazing!

  40. Mohit Nagarkoti on May 27, 2021 at 5:29 am

    Great video.

    I have question.. Can i practice Deep learning using Google colab/ kaggle notebooks?

  41. Data Professor on May 27, 2021 at 5:30 am

    Ken, Thanks for another quality video! This is a golden piece of advice on the million dollar question. The 3 tier starter specs are really useful starting point for building our own PC. Currently most of my work are down on my MacBook and run on the cloud. Will definitely, re-watch this video when I actually go and build my own PC.

  42. ali altฤฑntaลŸ on May 27, 2021 at 5:31 am

    What about linux or macos or windows? It would be amazing to make video adv and disadv of them in data science:)

  43. ๅฆ‚ๅฆ‚์—ฌ์—ฌ on May 27, 2021 at 5:34 am

    ์ผ„์ง€ ์งฑ!

  44. Jacob Gladman on May 27, 2021 at 5:35 am

    Have you considered doing a video discussing the different operating systems? Lots of conflicting opinions between Windows, Mac, and Linux and very few resources for data science specifically. Would be great to hear your opinion on this matter

  45. Vyas Vasudevan on May 27, 2021 at 5:35 am

    Hey SSD stands for Solid State Drive not steady state drive. Just wanted to let you know.

  46. Import Data on May 27, 2021 at 5:35 am

    The video I’ve been waiting for! I’ma save some money and go for the mid tier one as I wanna get back into gaming!

  47. Daniel Hernandez on May 27, 2021 at 5:37 am

    Hi Ken Jee! I have a macbook pro 16 2019 but i need it to work with Power BI and Power Query. I am struggling with that. I may probably need to buy another computer. What do you suggest?

  48. Koizer on May 27, 2021 at 5:37 am

    Hey, could you make a vid explaining how to get a gpu in the cloud in your laptop, I have all specs but my gpu is integrated.

  49. Ismail Muminov on May 27, 2021 at 5:38 am

    For beginners, go with what you have. Of course it is nice to have high specs, but to get started you dont need that much power. You can do pretty much anything on kaggle, googlw colab, or native jupyter notebook. If you get professional enough, you will need what kind of computer you need.

    For much of data cleaning and learning and linear regression, average computer is just fine. You dont need nvidia cuda cores for that. Waste of money for beginning.

  50. John Wade on May 27, 2021 at 5:40 am

    Tell me more about that usb-c graphics card tech for laptops.