Data exploration - Where my data engineers at

Charter / Eng
Baelfur

Charter Eng

PRE
T-Mobile
Baelfurmore
Jul 15, 2017 15 Comments

Telecom network engineer here, I got pigeon holed into data engineering responsibilities because as far as I am aware we don't have those. What are your favorite tools/ best practices for ripping into new data sets? Identifying underlying causality?

We recently were provided Tableau licenses, which work well enough, but I constantly run into scalability issues. Personally I prefer complete joining complete tables and pivoting/ visualizing them until I find underlying nuances, but seem to run face first into resource issues constantly.

Interested to hear others thoughts and opinions!

comments

Want to comment? LOG IN or SIGN UP
TOP 15 Comments
  • Tableau / Eng DAT-A
    I think we have an update coming you may like :p
    Jul 15, 2017 5
    • Tableau DikD20
      hyperhyper
      Jul 15, 2017
    • Tableau / Eng DAT-A
      I was referring to maestro but yeah... Hyper hyper!!
      Jul 15, 2017
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      Both hyper and Maestro sound amazing from what I've read so far.

      also, whoever thought up the name Maestro seriously needs to be given a bonus.

      'What do you do for work?'
      "Sir/ Madam, I conduct symphonies of information."
      *Mic drop*
      Jul 16, 2017
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      Also as a side thought for Hyper. Forgive me if this is a function of server and I need coffee, it would be really really cool if you could connect tableau to an external VM literally just for the hardware bank. Something where a user has a one time large job so they RC to a VM that has 128Gb of Ram/ 32 core I9 just cause. Do your work for the day to get your insights, then disconnect and move on. it would be a very interesting way to sidestep user hardware limitations.

      I hate asking questions that are to big for systems to handle. Makes me sad.
      Jul 16, 2017
    • Tableau / Eng DAT-A
      Sounds like we have several updates you'll like. There's VM integration work being done, not sure of the details or status on that.
      Jul 16, 2017
  • New / Eng brodriguez
    Tableau is really hard to beat for rapid, beautiful vis. If you want something the can scale at the cost of visual appeal, python is my go to, particularly if you have a spark cluster.
    Jul 15, 2017 2
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      I will definitely look into that, thanks!
      Jul 15, 2017
    • Databricks dumbdude
      Try Databricks community edition. Free spark cluster
      Jul 16, 2017
  • Tableau _______
    I feel like this is the first non negative/troll thread I've seen on blind :P
    Jul 15, 2017 1
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      There be smart people here, why not use them for their brains imo :P
      Jul 16, 2017
  • Facebook IBoW18
    looker is on the way to a leading data platform
    Jul 15, 2017 1
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      thanks, I'll give it a look!
      Jul 16, 2017
  • Proofpoint
    aLgQ81

    Proofpoint

    PRE
    Proofpoint
    aLgQ81more
    Depends on what you want.
    There is some talk I've seen about Dash - which is sort of like Shiny for Python. I haven't tried it yet but it might be something to look into for you.
    Jul 15, 2017 1
    • Charter / Eng
      Baelfur

      Charter Eng

      PRE
      T-Mobile
      Baelfurmore
      OP
      I really like adding 5-6 core pivots then adding 10-20 slicers to see how the distributions changes given different filter combinations.

      the main platform that I have ever used that delivers something close is Salesforce Wave.

      basically being able to scope in and out with dynamic specificity without having to manually build tables and rip them down over and over, but on a scale of 2-5 huge sets
      Jul 16, 2017