I'm late in reply, but for any other people looking for system for Data Analytics, please go with following configuration:
1) CPU: Preference for Intel CPU over AMD preferably 2nd or 3rd top version. Give Low Priority for overclocking
Reason: My experience with data software is that though most software itself are multi threaded, its the plugin/code that sometimes use single thread and thats when I find Intel better than AMD. For eg my most macros functions in Excel run in single core single thread, this I have never been able to fix in most cases. Similarly many of my R and Python plugins for data run on single core single thread.
SAS and SPSS do run ok as I mostly use its inbuilt functions.
2) RAM: 16 GB to start with, but go for 32 GB if possible. CAS latency is not a big issue, for data analytics high CAS latency and lower frequency RAM is ok. So you can buy the cheap RAM.
3) HDD: SSD nvme M2. Go for min 256 GB, 512 is better. I had issue in 128 GB as OS and core software took 45 GB. Rest space for data was low therefore was unable to move work data in fast drive. 512 GB made the whole change. It removed the bottleneck. Throw a 7200 RPM HDD as secondary disk. Transfer work data on SSD and then after work done transfer it back to HDD.
4) Graphic Card: Let me put this upfront, unless you are not doing coding in CUDA or OpenCL, graphic card is useless. Excel doesn't use it for calculations only graphs. R and Python also uses CPU and RAM and GPU is idle. So unless your software or code is capable of doing GPGPU, avoid investing on GPU, use that money for better cpu and ram. Or go for cheapest Pascal series nvidia GPU for minor GPGPU work and Turing series GPU for extensive work as it has dedicated Tensor core. Avoid AMD as nvidia is only card that support CUDA and opencl is supported by both. Also I have mostly come across GPGPU projects using CUDA/Nvidia.
5) Screen: As big you can go for. Since my visuals are mostly static or not very complex moving, I sometimes even connect to my TV with bigger screen instead for reduced scrolling.
Any other things will be add-on bonus but won't make much drastic changes to overall performance.
My configuration is based on following software that I have used:
SAS (base and enterprise), R (via rStudio), Python (via Anaconda), SPSS, Excel, Postgres Server (for local and remote database),
