Video ID: Tow21xoMhL0
YouTube URL: https://www.youtube.com/watch?v=Tow21xoMhL0
Added At: 13-06-25 21:17:21
Processed: No
Sentiment: Neutral
Categories: Science, Tech
Tags: data science, laptops, machine learning, AI, Excel, Tableau, PowerBI, Python, Matlab, Rust
Summary
The video discusses the best laptops for data scientists, including those doing machine learning and AI. The host shares their experience as a data scientist and interviews other professionals to determine what features are most important in a laptop. Key takeaways include the importance of display size, brightness, and resolution; comfortable keyboard; accurate trackpad; and sufficient storage and memory.
Transcript
today I'm going to cover the best laptops for data scientists including those doing machine learning and AI I'm so excited folks I've been wanted to do this video for some time and today I'll be debooting something special to make this guide as comprehensive as possible I interviewed several data scientists working in the field I'll be combining their feedback with my own personal experience as I did a lot of data science myself over the years anyway this guide is going to be applicable for students as well as professionals now to make this as easy to understand as I can I've come up with three representative data scientists let me introduce you to Lilia she is an investment banking analyst who spends her day going through company's financial statements to determine whether they are overvalued or undervalued to do this she primarily uses Excel for which she is a black belt ninja rat but for analytics that are too large for Excel she leverages business intelligence tools like Tableau and powerbi next up is Bob he is a quantitative data scientist and he works in the same bank as Lilia he spends his time building models that forecast interest rates and the price of Commodities that may impact the companies in lilia's portfolio he spends his Day writing code in Python mat lab andr that executes on remote servers and then we have Agnes she works on a large language model for the interior design company decorilla her model is used by their chatbot which listens to client's design needs and then responds with ideas she tests and refines her models locally on her laptop before putting them into production on the company's Cloud servers she works in Python and rust while leveraging libraries like pytorch now taking a look at these three personas from a laptop buying perspective Bob and lilia's needs are actually somewhat similar Lilia is only doing basic analytics on a laptop Bob although a quantitative does not execute his models on his laptop instead they execute on a remote server so similar to Lilia his actual performance and data storage needs on his laptop itself are basic therefore I'm going to first cover what kind of laptop data scientist like Lilia and Bob need and then I'm going to build on that with some additional needs for a data scientist like Agnes who has much higher Computing needs if you want to just jump to that section feel free I won't be offended for data scientists you'll be working with a lot of data and graphs or even code you want to see enough of them as possible on screen to really know what's going on therefore unless you're always using your laptop plugged into an external monitor you want a display that is 15 in or larger but it's not just the display size that determines how much you can see on it it's also its brightness and resolution I'd recommend a screen with a minimum of 400 nits of brightness but preferably 500 this will enable you to see the display clearly in a variety of environments without needing to squint brighter environment such as rooms with direct sunlight will require your laptop screen to be brighter and that display certainly helps with this as it is non-reflective but it won't completely solve this issue and when it comes to the display's resolution I recommend around 220 pixels per inch or higher this will make small text and numbers just easier to make out as they will appear crisper a fast refresh rate display is also a nice bonus as it will make scrolling through lots of data just look silky smooth when it comes to the laptop's chassis itself many data scientists don't work alone they frequently demonstrate their work to business partners and other team members having a laptop that is light enough to easily pick up and take with you is ideal I'd try to get one that is under £5 but preferably under four this this will also make working in environments like coffee stores or co-working spaces just more pleasant now for the keyboard you'll want one with a comfortable satisfying click and a standard layout there is nothing worse than typing on a keyboard that feels uncomfortable and hurts your wrist or reaching for a key and pressing the wrong one on this note if you're going to use a lot of excel like Lilia you'll want a Windows laptop the hot keys for Excel are different than on Mac OSS and you just won't be as efficient also something to watch out for is whether the keyboard has a number pad some data scientists really like them and others don't it's a personal preference but just be aware that getting a number pad isn't free it forces the laptop's keyboard to be off center which you may not like for the trackpad you want one that is nice and accurate so you can quickly do things like click to a data cell look for one with a haptic trackpad those are the most accurate now for the processor inside the laptop it's become quite complex of late with a ton of new ones announced for data scientists like Bob their Computing needs on their laptop itself are light as his processing as I said it's all done in the cloud for Lilia it depends on how large her data set is and what operation she is performing on it if you believe the compute needs of your laptop are light all modern laptop processors from Intel AMD and Apple sold in new laptops $500 and up are good enough if you believe your Computing needs maybe more look for a laptop with an Intel Core Ultra 9 processor or an Intel hx1 you could also get an AMD Zen 5 processor or one with an apple Pro or Max chip quick note of caution on a couple of specific processors Qualcomm Snapdragon processors use the arm version of Windows several data science programs like Tableau powerbi and matlb they do not run natively on this Hardware I tested them for this video Tableau and powerbi they do run well under emulation but matlb does not regardless you will never get the full performance out of the hardware that you are buying for applications that run via emulation so only get a Snapdragon powered laptop if you are certain that the programs you are running will work Excel is a good example of this now on powerbi specifically the desktop version doesn't run on Mac OS so keep that in mind finally do not buy a Chromebook for data science you'll be even more limited on what applications you can run now in terms of the configuration of your laptop I'd like to see you with a minimum of 16 G of memory and 512 gig of storage but I'd strongly recommend more if you can afford it 32 gig of memory and 1 terab of storage is a good safe amount memory and storage are very important in data science fundamentally you're working with lots of data and that takes up a lot of space and when performing operations on it it needs to be put into memory which will use up a lot of memory obviously the amount depends entirely on the data set you are working on and what kinds of analysis you are doing on your laptop itself so these are General guides let's talk battery life there have been huge improvements in laptops with the latest generation of processors this includes the following ones any Apple laptop with an M series processor intels lunar L processors also known as core Ultra 2 AMD N5 processors and Qualcomm Snapdragon X processors all of these will give you fairly decent battery life so there is more of a benefit to buying a laptop with a newly released processor than there has been in Prior years that being said I personally carry this tiny ug green nexo pro 65w charger with me no matter what laptop I'm using it delivers enough power to drive most laptops while also charging your phone this video is not sponsored by ug Green in fact even for the sponsored spots that we've done for them they don't seem to want me to promote this charger I have no clue why as it really is a Marvel of engineering anyway I thought I'd mention it as I don't think battery life should be make or break when such a portable charger is available in addition to the attributes that I've already mentioned when you're in the zone you want to minimize distractions this means that you want a laptop that doesn't feel overly warm to the touch and if you're in a quiet environment you won't want to be distracted by fan noise in general we've found laptops with an apple M series Intel Luna Lake core Ultra 2 or Snapdragon X processor are s tier for minimal heat and fan noise laptops with an amd's N5 processor are a tier and they are followed by amd's older N4 processors in Intel's core Ultram media Lake which are B tier Windows laptops that are older than this will be noticeably louder and feel warmer now we're going to turn it up a notch when it comes to training a model many professionals and researchers like Agnes do test their models on their laptops there are numerous reasons for this one such is that server time can be very expensive especially on supercomputers you don't want to waste their processing Cycles by running a model that may not work therefore it is valuable to have a laptop that is powerful enough to at least test your model on and by the way I'm referring to testing the training of a model that is more intensive than model inference which is using an already trained model all right when it comes to that training of a model gpus are more powerful than CPUs that's because they are better at executing Matrix operations which are the back bone of machine learning and artificial intelligence now when choosing which GPU keep the following in mind nvidia's Cuda API is the most developed instruction set it of course only works on their gpus now you can use amd's dedicated laptop gpus as well as the integrated gpus in MacBooks but they're just not going to be as industry standard nor as powerful as nvidia's and in addition to which GPU to get memory is critical to how well your models run there are up to two types of of memory in your laptop Graphics memory which runs fast and is only available to your dedicated GPU and then there is system memory which is used by your CPU integrated GPU or mpu it runs slower but you can get more of it having enough memory to fit all the data required to test your model is vital dedicated gpus in laptops they have limited memory available currently between 6 gig and 16 gig so keep that in mind when picking your laptop now if you absolutely need more mem than what's available in a laptop's GPU or at least one you can afford you can indeed run machine learning tasks off your laptop's CPU but your model just won't run nearly as fast one interesting exception to this is Apple's MacBook Pros their integrated gpus are extremely powerful and since their memory is Unified you get the benefits of faster speeds similar to Graphics memory and you get access to more memory up to 128 gigs which is similar to the benefits of system memory so it is totally viable to choose an Apple laptop for machine learning for that reason even if it means missing out on Cuda by the way you may have heard about powerful mpus in the latest generation of laptops please be careful here these are exciting for sure but right now their Terror operations per second are way behind even the slowest dedicated gpus and since they use system memory they're slower again another heads up AMD is coming out with a new set of laptop processors called STS Halo these promise to do something similar to Apple combining a high powerered CPU with an extremely powerful integrated GPU that can share memory all right time to bring all these learnings together into recommendations for buyers like Agnes I'd recommend getting a laptop with a dedicated Nvidia GPU that has as much Graphics memory as you can afford however having enough system memory is also important it's a nice backup if you can't fit all your data in the gpus memory and you want to execute via your CPU when it comes to system memory I would recommend a minimum of 32 gig but ideally 64 the other path you could of course take is to get a MacBook Pro for the reasons that I've already talked about with this all said it's now time to recommend some specific laptops these are laptops that come as close as possible to achieving the recommendations I just made none of them are perfect links to the ones that I talk about will be in the video's description and on our website in fact it is very important that you visit the best laptops for data scientist lists on our website just josh. Tech new laptops they come out regularly and that list will be kept up to date well after this video goes live also this is a best laptop video these ones are therefore pretty pricey I will put a selection of Cheaper laptops on our website as I do want to keep this video actually short enough to be watchable and on that note I'm only going to hit on the highlights of each one you'll find links to our full reviews of each of the laptops I mentioned on the website too we're going to start with a yoga slim 7i or Edition it is a very light laptop for for one that has a large 15.3 in display it screen is excellent for data science work bright high resolution fast refresh rate and the laptop has an extremely comfortable keyboard it comes with Intel's latest lunar Lake processor so it gives you very long battery life no hidden fan noise and as it's an Intel processor the broadest application compatibility also if you want to do some gaming in your downtime this processor has the most powerful integrated Graphics that we've tested in a thinner light Windows laptop now even though this processor does have excellent single core and integrated GPU performance its downside is that it has poor multi-core performance so it's not the most powerful of laptops I recommend it for data scientists with basic performance needs of their laptop itself like Lilia and Bob my next recommendation is the MacBook air15 just like the yoga slim 7i AA Edition this is another lightweight laptop with a large 15.3 in display it's actually easier to pick up and carry around than the slim 7i Aura as the air is thinner compared to that laptop the MacBook Air giv you comparable multi-core performance faster single core performance and better battery life that being said it's a similar class of machine to the aura in terms of its overall performance which is basic so again suitable for data scientists like Lilia and Bob the biggest difference between the air and the aura though is in Creature Comforts The air15 it's a more all round premium laptop so the little things on it like the webcam and speakers they are noticeably better than on the yoga on the flip side though the yoga has a fast refresh rate display and a more comfortable keyboard many people including me will prefer the aura because of that the proot p16 is a great option for data scientists who are looking for performance yet they still want a laptop that is somewhat portable the proart uses amd's brand new Zen 5 stck point processor the laptop performs very well and it is effectively cooled so it doesn't get warm to the touch nor have annoying fan noise it is very light and portable for a 16inch performance laptop and there are a variety of different configurations available at the bottom end you can get it with 32 G of memory 1 tab of storage which is pretty good and an Nvidia RTX 4060 but you can spec it right up to 64 gig of memory 2 terab of storage and an RTX 4070 with 8 gigs of Graphics Ram it also has an incredibly crisp 4K plus display it's main downside though is about that display while stunning it doesn't get that bright and it doesn't have a fast refresh rate if this annoys you a great alternative is the yoga Pro 9i 16 in it's mini LED panel is high resolution bright and has a fast refresh rate its main downsides are that it is heavier and it looks more dated than the pro art also for some reason in the USA it can only be had with up to an RTX 4060 and 32 gig of memory elsewhere in the world you can configure it with an RTX 4070 and 64 gig of memory all right now we're going to step it up to the final Notch and talk about very powerful laptops that are going to be the best out there for doing ML and AI on your laptop itself just be prepared these big boys are expensive I'm going to start with the MacBook Pro 16 I'm not referring to the Basse models I'm referring to an upgraded configuration with a Max chip and plenty of memory the max chips have significantly more powerful gpus and higher memory bandwidth as mentioned earlier in this video this is a rare laptop that can combine a powerful GPU with up to 128 gig of very fast memory exactly what ML and AI model training needs other than this the MacBook Pro 16 is just a fantastic around machine I personally have chosen to use it as my high performance laptop it has it all it's well built it weighs less than most other high-powered laptops it has great battery life a good keyboard the best trackpad fantastic speakers you guys get the point that being said it's a Mac so it is super expensive and you don't have Nvidia Graphics so you don't get Cuda and most AAA games aren't available for it now if you want something super powerful that runs windows or Linux and has an Nvidia Graphics the legion Pro 7i it's a great pick this laptop has a very powerful Intel 14900 HX processor that is fed a ton of power so it performs very well it can be hand with up to an Nvidia RTX 490 with 16 G memory and the laptop itself is very well cooled which again it helps it perform extremely well it has two upgradeable memory slots so you can spec it right up to 64 gig of system memory it also has a great screen and a comfortable keyboard its main downsides though are that its displays pixel density isn't quite as high as I'd like and it is quite a thick laptop but given the performance of the components inside what can you expect now if you want The Most Extreme Performance in a 16-in laptop the electronics hydrock it's the way to go it gives you the same components as the legion Pro 7i but it adds the option of water cooling those components for even more performance it also uses rare 48 gig memory chips so you can spec it up to a whopping 96 gig of system memory if you're not in the USA this laptop is sold by other companies under different names check out xmg if you're in Europe or Aftershock if you are in Australia finally you could go the 18-in laptop route with the MSI Titan 18 that laptop actually has four slots for system memory so you can spec it up with 128 gigs but at this point given its enormous weight and price what are you doing you really should be buying a desktop well that's all for today folks as mentioned we'll have more options in our best laptop for data scientist list on our website including cheaper ones in fact our website not only lists all the laptops that we recommend for a variety of different types of uses but it also shows you where to go to get the cheapest prices on them at least right now in the USA but in the future we will be expanding that feature to support viewers in additional countries also if you don't already know I do have a second Channel That's all about how to be successful in your career and in business before I became a full-time YouTuber I rose up the ranks of corporate America to the highest titled level you can get at a bank hopefully I can help you achieve your goals even faster than I did my business channel is linked below so go check that out YouTube is Dick time smash the like button and get subscribed not only does it help us grow which means that we can make more videos for you but as I always say it makes my dearest mother very proud till next time go do something awesome with your day and I will catch you later