22

Proposal: Data Science

Isn't the subject of this proposal already covered in

9

Well this is more of a place for Data Scientists/Machine Learning Engineers/Information Retrieval Engineers/Data Engineers and so on to hover on. Stats-Stackexchange (purely for Statisticians) does not cover anything on Mahout, Hadoop, Twitter Storm, scikit-learn, Solr/Lucene, Nutch/scrapy and endless list of such Algorithms/Softwares that are used in solving real world Data Science problems. It is now not just about Regression & R (stats.stackexchange), nor is it just about "I am getting a memory leak in my Weka program given below (stackoverflow.com)" there is much more to Data Science in the so called world of Big & Fast Data.

A sample question (which is not fit for all the sites listed by you):

How can I train Baum Welch Algorithm used in Hidden Markov Model in Mahout in parallel? 
  • 14
    " Stats-Stackexchange (purely for Statisticians) does not cover anything on Mahout, Hadoop, Twitter Storm, scikit-learn, Solr/Lucene, Nutch/scrapy..." That's because software implementation is the domain of StackOverflow. CrossValidated tries to focus on the theory behind applications. There's really nothing in data science that doesn't fit into either SO or CV (or dba). If you don't think that's true, you haven't participated in those communities enough. – David Marx Mar 13 '14 at 5:39
  • 1
    @DavidMarx What about the following questions (I am pretty sure they all will be closed on SO) 1. "How can I train Baum Welch Algorithm used in Hidden Markov Model in Mahout in parallel?", 2."I'd like to run multiple instances of Scikit-Learn in parallel on somekind of cloud computing service. What is the recommended way to do this?" 3. "What are the biggest pitfalls to avoid when using Map/Reduce to query data?" 4. "What is the difference between Twitter Storm & Esper?" and so on.. – Yavar Mar 14 '14 at 9:46
  • 6
    like I said: implementation questions go on SO. All of those questions are appropriate for SO. In fact, very similar questions have all been asked on SO already, none have been closed prematurely, and all but the first have received answers: stackoverflow.com/questions/19975520/… stackoverflow.com/questions/21052050/… stackoverflow.com/questions/28982/… stackoverflow.com/questions/619846/alternative-to-esper – David Marx Mar 14 '14 at 14:32
  • 6
    So the data science QA would just be about using existing software? - what's wrong with stack overflow? You might be surprised but many of the questions on cross validated are by scientists. That is 'real world' scientists answering research questions. As david marx says cross validated is suitable for the majority of theory related questions. I.e. what exactly are you doing with mahout/whatever (aside from dropping names of hot technology to look cool)? – conjectures Mar 15 '14 at 15:52
  • @DavidMarx - I couldn't get your point finally :) Do you think this site overlaps with SO and CV or not? – mok Mar 25 '14 at 20:26
  • 7
    @mok - My opinion is that there is very unlikely to be content in this stackexchange that does not belong on either SO or CV, and this particular community is not necessary as the niche it is trying to fill is already being adequately served by SO and CV. – David Marx Mar 25 '14 at 21:01
  • 1
    @DavidMarx although I'm interested in this proposal... I think you make a good point about questions being still a good fit for another site such as SO – Crowie Mar 27 '14 at 12:23
  • 6
    You're wrong about CV. There are tags for scikit-learn and rhadoop, and almost 2K questions on machine learning. If this site opens, you should probably plan to migrate questions about regression to CV, where the current count of such questions is 4316. They will not be better served on a beta site if they're primarily about regression. – Nick Stauner Apr 13 '14 at 20:28
  • 2
    Apparently suggested edits aren't allowed here, but to the author: if you can edit, I think the question would be improved to include each site's title and one-line description from its tour page. For example, "Cross Validated: a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization." – xan Apr 15 '14 at 13:23
  • 1
    "Stats-Stackexchange (purely for Statisticians)" - this comes as a surprise to me, a CV user with 15K rep who is most definitively not a statistician. – Fomite Aug 15 '17 at 20:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .