onsdag 24. august 2016

Laptop for data science

What are the laptops which are most suited for data scientists and analysts?

As we deal with heavy computations and also need to generate visualizations, something which can take the toll of it, would be recommended.

Would be preferred if it can help in handling Big Data analytics too.

Even though the analytics is done in the Map Reduce framework (or distributed computing), yet the computations are heavy and time taking and also slows down the laptop in most cases.

So, a laptop with features and OS which is most suited to handle such things gracefully is recommended.

[Price not an issue]

As I am pretty much in the same situation, here are what I look for:

SSD: since you'll likely perform many I/O on large data sets. 1 TB is my bottom line.
RAM: since it's often more convenient and much faster to keep data sets (or part of it) in memory. 16 GB is really bottom line.
GPU: Nvidia is sometimes preferable over AMD as it tends to be more supported (e.g. for neural network libraries). I had to get a MBP M2014 instead of M2015 because the latter had AMD while the former has Nvidia, and I need to use Theano.
OS: Linux tend to have more libraries (but since it doesn't have any decent speech engine software I personally use Microsoft Windows, using Linux in VM or in server).
CPU: hasn't evolved much over the last few years... some i7 3rd or 4th generation is standard.

As it's often cheaper to add SSD and RAM oneself, I tend upgrade mid-spec laptops.

If price isn't an issue, you can have a look at those overpriced Alienwares. For people who are more budget conscious, just check to what extend the laptop is upgradable (e.g. max RAM + number of SSD slots). In the US, I like Xotic PC as the max specs are clearly defined.

Ingen kommentarer:

Legg inn en kommentar