New (late 2020) macbook air m1 for epidemiology PhD? To go for it or no?
February 19, 2021 9:37 AM   Subscribe

I am in the market for a new laptop in anticipation of starting a PhD program in epidemiology and am trying to decide whether to take the plunge with one of Apple's M1 Macbook Airs. Anyone out there running relevant software have feedback? Much appreciated.

Relevant info:
1. Have been using macs as PCs for my entire adult life. Have Dell for work, far prefer my mac
2. Upgrades I plan to get if buying an m1 machine: 512 GB memory, 16 core CPU. I also have a 2 TB Synology NAS I can use to supplement.
3. Software and programs I will need to use: R, SAS, STATA, possibly SPSS, ArcGIS and other GIS software (I need to learn some of these in case you are wondering why I don't just test them out myself). And I plan to install MS Office Suite.
4. I know I'll also need to get more familiar with Python and SQL... thinking command line access on mac can help with this.
5. For various reasons, waiting for the m1's next iteration is not an option.
6. Am not a programmer and a data science newbie prepping for a steep learning curve :)
7. I see good reviews online for the m1 for data science generally but am not finding reviews more specific to the kinds of statistical/GIS packages I will be using.

Any feedback from those who use this sotware for statistics/epidemiology or related field especially appreciated.

Thanks, hivemind!!
posted by dubhemerak3000 to Computers & Internet (14 answers total) 2 users marked this as a favorite
 
I use R, Python, and MS Office on my m1 MacBook without issue. I can access databases (Postgres, MySQL, redshift) without issue, but haven’t tried running the databases locally. I don’t have any plugins in Excel, and I don’t use the rest of the software you that you mentioned, so can’t comment on them.

So far the only application that isn’t working on my m1 Mac (Fidelity’s ActiveTrader Pro) was also terrible on my Intel Mac. It just switched from crashing a lot to crashing every time.
posted by whisk(e)y neat at 9:45 AM on February 19 [1 favorite]


While I spend the most time infant of an iPad-as-terminal I did recently replace a MacBook Air with an M1 Mac mini; spec'ed similar to your description.

I spend a lot of time crunching with python, a little with R, and a stack with a whole host of tools. For reasons I'm currently running most of this with the help of Rosetta 2 emulation. The machine is still 3-4x faster for my purposes than the (2017) MacBook Air it replaced and nearly twice as fast as the headless linux box in the closet.

And cheaper than replacing either one of them. Personal experience says you'll be fine. Unless you need great support form Adobe now, or are a sound engineer, these machines are what most people should be buying now.
posted by mce at 10:16 AM on February 19 [2 favorites]


Do note that ArcGIS is only available for Windows, and the M1 cannot (currently) run Windows, even in emulation. You could run it on your Dell and use Remote Desktop to get into it from your Mac, however.

But the strictly *nix and macOS tools should work very well, and very fast, on even the lower-end M1 Macs.
posted by eafarris at 10:21 AM on February 19 [3 favorites]


Oh that's an excellent catch safaris. And given the recommended setup (https://pro.arcgis.com/en/pro-app/latest/get-started/run-pro-on-a-mac.htm) it's not likely to work with parallels technology preview which only runs windows for arm.
posted by mce at 10:50 AM on February 19 [1 favorite]


Windows for ARM does include x86 emulation, so ArcGis might work under Parallels—still pre-release software, though.

If you want a Mac, definitely go M1. The performance is amazing, and the transition from Intel has been remarkably smooth.
posted by vitout at 10:55 AM on February 19 [1 favorite]


When I took my GIS & Public Health course a few years back with a MacBook Pro (not an M1) I used QGIS as an alternative to ArcGIS and while there were quirks, it worked well enough. I had to spend some extra time with him after class ironing out some kinks, but it was fine for my purposes.
posted by Ufez Jones at 11:01 AM on February 19 [4 favorites]


Whatever you get be sure to check if the university has some discount agreement with Apple.
posted by mareli at 11:03 AM on February 19 [2 favorites]


If you need to run Windows x86 applications without emulation or via remote desktop, your safest spend is still an Intel-based Mac, unfortunately. However, the highest-spec'ed Intel-based MacBook Pros are still just on par with the M1 in performance, albeit somewhat more expensive and noisy/hot. If you can go without Windows applications, an M1 is among the best portable data science tools you can get.
posted by They sucked his brains out! at 11:15 AM on February 19 [1 favorite]


I'd be nervous about it. ArcGIS is going to be a big limiter as mentioned above.

Last time I looked, a lot of people were using the educational release of SAS in Docker which currently doesn't run on the M1. The R Studio docker images are also far easier to setup and run locally. It's possible that your program would have RStudio, SAS, and the like available to you on a server so you could use those through a web browser. It's one of the best things to happen to stats software in a while.

An Intel Mac would let you dual boot and run ArcGIS in native Windows. You'd probably be fine with all the other stuff on the Mac side. I have the latest Intel MacBook Pro and I really don't feel like it's going to be a limiter to me anytime sooner than a new M1 would be.

I don't use any of this software, but I do manage the server team for a demographic data institute that has a lot of faculty and grad RAs in epi.
posted by advicepig at 1:14 PM on February 19 [1 favorite]


Seconding QGIS as a possible alternative to ArcGIS. I used QGIS open source application 8 -- 3 years ago to view and edit lots of geospatial data, mostly ESRI shapefiles but also geospatial data stored inside postgresql DB with postgis, running QGIS under both windows and linux and on standard X86-64 architecture machines. QGIS is implemented using lots of python and c/c++ open source geospatial related projects, so if you can get QGIS to run that's a good sign that lots of open source python geospatial libraries and lower level libraries like GDAL and GEOS will work as well.

There is evidence that at least one person in the world can successfully run QGIS to view spatial data on apple m1 silicon hardware. I believe QGIS only runs in compatibility mode after being automatically rosetta 2 translated. In any case, a positive sign!

Allegedly QGIS has already been ported to run on android (ARM CPU architecture, similar/same as apple M1 (?)), so perhaps in a few months or few years it will also be possible to download a build of QGIS that can run natively on apple M1 silicon without needing rosetta 2 translation.
posted by are-coral-made at 2:14 PM on February 19 [1 favorite]


I run ArcMap via an emulator back to my Mac, and it runs fine, but I hate doing it because even on the big ass MacBook Pro screen, it’s really small, especially if I’m also having to set up a data file at the same time. It’s the one thing I will truck into campus and use the windows labs with their double monitors for.

So if you’re going to have a computer lab with GIS tools available to you, and you can plan your time such that you do those tasks on campus, then you should be fine.
posted by joycehealy at 4:30 PM on February 19 [1 favorite]


SAS and ArcGIS will not run on a Mac. I would strongly recommend using a windows machine for epi computing. QGIS is great, but it’s not taught as often and it does have gaps specifically for epi/stats stuff. You can get around it, but your classes will be taught with Arc and your TAs won’t be able to translate.

I bought a Mac for my MPH and really regretted it. I ran SAS and Arc in Parallels, and it was super laggy and messy (Arc is often laggy and messy; the combination was frustrating). I ended up doing all my coding and mapping at the school library, which wasn’t ideal.
posted by quadrilaterals at 9:49 PM on February 19 [1 favorite]


Also, R on a Mac is fine 85% of the time. The other 15% are really frustrating edge cases that never get resolved. That might be specific to the type of work and data presentations I do, though.
posted by quadrilaterals at 9:52 PM on February 19 [1 favorite]


Response by poster: Thanks so much, all, for all this great food for thought. An update for anyone else that might be in a similar boat - some developments on Docker - https://www.youtube.com/watch?v=wsg7wrTbBRk

Note, though, that it's not fully up to speed yet
posted by dubhemerak3000 at 12:46 PM on February 23


« Older Postcard Printer or similar   |   Take it down a notch (Voltage edition) Newer »

You are not logged in, either login or create an account to post comments