Subscribe to DSC Newsletter

What home computer platform do you recommend for a data science practitioner?

Dr. Granville recommended that I post this message here on the forum.

I am in the market for a new notebook PC. Can you recommend one to me, your preferred brands, and the minimum system dimensions you feel are important for a data scientist to have on their working platform? Are Apple products satisfactory for serious business users? I have always worked on Unix variants and Windows platforms for systems and software engineering work.

I intend for a notebook PC to give demonstrations of my unique value proposition as a data scientist when I solicit new business during personal presentations. I also intend to rely on it for graphically demonstrating my work products with data visualizations, modeling and simulations. I may experiment with in-memory data management and predictive analytics. I intend to use such open source tools as VisIt, Gephi or ParaView. Do you recommend others?

I will likely keep production-level big data in the Cloud or on my clients' own data platforms, but I will nevertheless still be processing some large data sources on my local system, depending on what my clientele give me for customer experience, financial markets, macro-economics, industry studies, or scientific and engineering analyses.

Thank you very much for your consideration, and best wishes always.

Tags: PC, UI, computer, hardware, interface, laptop, memory, notebook, platform, storage, More…user

Views: 4854

Reply to This

Replies to This Discussion

It isn't going to matter a whole lot in the end. All processing barriers you run into can be resolved with an EC2 instance or similar. I work for a mobile gaming company with data from millions of players across many titles and my work computer is a run-of-the-mill Macbook Pro. All of my data work so far has been easily done by pushing queries on data in Redshift to S3 and pulling that down to my local machine. Some models can take awhile to run, so I just benchmark on a random sample and train overnight. I have yet to run into any serious processing roadblocks and if I did we could easily just fire up an EC2 instance.

My home computer is a Gaming system that I built myself, and it is far more than I would ever need for Data Science work.

Thank you very much, Christie. This is practical advice. I am just learning about AWS and how I might incorporate Redshift in my architectural planning.

The first thing I'd suggest to focus on is RAM memory size - more memory is always better for analytics and machine learning. Here is my related blog post "How Much Memory Does A Data Scientist Need?" and a discussion in the comments: https://fullstackml.com/2015/12/06/how-much-memory-does-a-data-scie...

Next is hard drive and CPU - depends on your area and data. Based on my experience, it is much easier to overcome HDD\CPU limitations than a lack of RAM memory.

The third is GPU. If you work with deep neural network or just NN you need a good graphic card.

Mac vs. Windows vs. Linux - it depends on industry\company you work for or your personal preferences. I don't see a difference.

And number zero is a screen :)

Thank you very much, Dmitry. Yours is all excellent advice I will follow during my shopping.  :)

I will also always prefer the higher clock-speed of the board. Very important.



Dmitry L. Petrov said:

The first thing I'd suggest to focus on is RAM memory size - more memory is always better for analytics and machine learning. Here is my related blog post "How Much Memory Does A Data Scientist Need?" and a discussion in the comments: https://fullstackml.com/2015/12/06/how-much-memory-does-a-data-scie...

Next is hard drive and CPU - depends on your area and data. Based on my experience, it is much easier to overcome HDD\CPU limitations than a lack of RAM memory.

The third is GPU. If you work with deep neural network or just NN you need a good graphic card.

Mac vs. Windows vs. Linux - it depends on industry\company you work for or your personal preferences. I don't see a difference.

And number zero is a screen :)

useful explanation, thanks!  top end Mac pro satisfies all other requirements, but how good is the GPU in Mac pro for deep learning? CUDA works only with NVDIA GPUs which are not available in Mac laptops! Is there a way to make GPUs in Mac pro work for deep learning?

Regards


Dmitry Petrov said:

The first thing I'd suggest to focus on is RAM memory size - more memory is always better for analytics and machine learning. Here is my related blog post "How Much Memory Does A Data Scientist Need?" and a discussion in the comments: https://fullstackml.com/2015/12/06/how-much-memory-does-a-data-scie...

Next is hard drive and CPU - depends on your area and data. Based on my experience, it is much easier to overcome HDD\CPU limitations than a lack of RAM memory.

The third is GPU. If you work with deep neural network or just NN you need a good graphic card.

Mac vs. Windows vs. Linux - it depends on industry\company you work for or your personal preferences. I don't see a difference.

And number zero is a screen :)

As far as I know - there is no way to use CUDA on Mac Book Pro properly.

I personally, did not buy the GPU version of Mac Book Pro for this reason. But it was 2 years ago - I don't know much about new Mac laptops.

RSS

Videos

  • Add Videos
  • View All

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service