What is data science?

Oracle 3not3
Aug 31, 2018 29 Comments

When did statistical analysis and curve fitting became known as data science? And who is a data scientist? What do these “scientists” do?

comments

Want to comment? LOG IN or SIGN UP
TOP 29 Comments
  • New Dtj72n
    If you think curve fitting is easy, you know nothing about it.

    Understanding business problems, gathering data, cleaning it up (toughest job ever), merging datasets which have conflicting info, fuzzy matching them, creating master datasets, exploratory analysis, curve fitting, model evaluation and improvement iterations, residual understanding and partial probability computation, statistical significance, and then converting everything to layman terms so software developers/business teams can understand and deploy those models in place.

    This is what data scientists do. Yes, I do this every single day.
    Aug 31, 2018 8
    • New Dtj72n
      Thanks @Microsoft. Been doing this several years now, not sure what the companies are looking that I can't get through the interviews. Looks like I just gotta keep trying.
      Aug 31, 2018
    • Microsoft qqxS28
      Yeah it sounds like you absolutely have the chops and know-how to be a strong performer in any of the bigger names. Just keep interviewing, you'll crack through sooner or later, you seem like you know your stuff. Good luck!
      Aug 31, 2018
    • Square Jqty44
      There is a lot of noise with interviews, not to mention companies are not always clear about what they really want which leads to some werid interviews. Just keep applying.
      Aug 31, 2018
    • Uber hGFT63
      This is all run of the mill stuff that can be picked up with a quick read of ISL. Any self respecting engineer can do this. Data scientists should be busy with the hard problems — recent research, unique optimization problems, CNNs, etc. You def don’t need a PHD for boring ISL and a jupyter notebook.
      Sep 1, 2018
    • New Dtj72n
      @uber - read my answer again, esp the 'curve fitting, model evaluation and improvement...' part.

      Every curve fitting is an optimization problem at it's core. And CNN/RNN/RF etc all are different kinds of models.
      Sep 1, 2018
  • Amazon Dr. Savage
    Data scientists do models. Duh.
    Aug 31, 2018 7
    • Oath / Eng [object
      Dat curve
      Aug 31, 2018
    • Intel Path2Dirty
      We talking Victoria's Secret type models or Lane Bryant?
      Aug 31, 2018
    • Amazon Dr. Savage
      All models. Many do multiple models together in one night.
      Aug 31, 2018
    • Microsoft UMbR31
      They must have a lot of stamina to do multiple models
      Aug 31, 2018
    • Amazon Dr. Savage
      They train models so they have a lot of practice.
      Aug 31, 2018
  • Capital One Barry42
    Data Science is over rated, very few folks in the industry know about actual data science, Rest all is NOISE....most of the data scientists speak lot of theory and software engineering but when they actually implement they freaking write a set of rules and call this shit a Model which is ridiculous, no mention about statistical analysis, decision tree analysis or predictive modeling.
    Aug 31, 2018 4
    • Uber UrbanLiar
      Doesn't Capital One have a great DS org? Clearly best in class and tons of conference presence 😂
      Aug 31, 2018
    • Capital One Barry42
      Noise is all around, I don’t see data scientists doing actual data science...!!
      Aug 31, 2018
    • New Dtj72n
      What do they do at cap one, just curious. Not what I described in my comment above?
      Aug 31, 2018
    • Capital One Barry42
      Few say they are data scientists but spend most of time preaching software engineering to tech practitioners, I don’t see real data science happening yet...I have also observed the notion of high bar for data science, high bar should be across the organization...reason I say this, few DS don’t even have awareness of building their features with the appropriate data pipelines...my strong opinion DS and ML are over rated and there are no proper use cases and leadership to run!!!C1 is hiring a ton DS, if interested apply get a better or good TC and run away in couple years!!!
      Aug 31, 2018
  • Microsoft antimony
    "What is mathematics? When did slopes and sums become a viable career path lmao"

    - You right now, sounding extremely idiotic
    Aug 31, 2018 1
    • Oracle 3not3
      OP
      Oh I know what they do. But it’s funny that they had to rebrand them to sound cool. Two years ago nowhere there’s a data science post. Now it’s everywhere. That’s what I don’t get. It’s not like we just invented data science. It’s called statistics.
      Aug 31, 2018
  • New / Other
    simplolol9

    New Other

    PRE
    500 Startups
    simplolol9more
    Really depends on the company. At some places that's just a glorified analyst who can't code and does excel and tableau. At others, it's a software engineer who knows ML very well. Can also be anything between
    Sep 1, 2018 0
  • KPMG / Mgmt
    Bad_Kitty

    KPMG Mgmt

    PRE
    LinkedIn, McKinsey
    Bad_Kittymore
    Analysts who know R are called Data Scientists these days. :) Most of them don't understand the business or operational context
    Sep 1, 2018 1
    • Intuit GTO
      Not surprising coming from a consultant. I find your generic statement that entirely lacks context and is the sort of broad sweeping non actionable statement consistent wit what you call work.
      Nov 8, 2018
  • Bsquare KYha84
    Machine Learning one of the major data science activities was coined almost 60 years ago, what has changed is the availability of low cost computing and great community driven / open source software making the the field accessible to anyone with access to a computer. Every time I talk with the Data scientists I am amazed by what they are able to glean from datasets and the how powerful the predictive models are.
    Aug 31, 2018 0
  • Capital One Barry42
    Most of the folks speak about automated refits but have no freaking idea in terms of implementation....
    Aug 31, 2018 0