1
$\begingroup$

I love that people are coming here to answer some of them, and I think we should encourage that for basic questions, and try to emphasize the conceptual aspects, and point to Data Science for more advanced followups, but

We need to start migrating new DS questions to DS

None of us AI mods have mod privileges on DS, which is why I, at least, have been reluctant to migrate.

  • At present, the "data-science" tag has only 11 instances on SE:AI

We need Data Science users with high rep to start tagging the questions to be migrated. If you're a trusted member of that community, I will migrate the questions for you.

We currently have only 11 data-science tags, with 3 unanswered:

Performance Evaluation Metrics used in Training, Validation and Testing

the above one I actually like for AI because it's asking about the concepts. I'd want to see the answer, ideally with links to Data Science questions.

Deep NN architecture for predicting a matrix from two matrices

Forecasting and predict using matlab Artificial Neural Network

Should we try to get these answered, and point to DS?

$\endgroup$
  • $\begingroup$ Extremely similar past discussion: Should we do a mass migration of Data Sciency questions on AI that remain unanswered? $\endgroup$ – Ben N Jun 2 '18 at 3:05
  • 1
    $\begingroup$ Thanks for linking to previous threads! I modified this post to reflect the reality. I'd definitely be willing to give a few migrations a try, because there have been some vocal AI users. I want to work with the related stacks, and am hoping that will encourage power users from those stacks also to participate here. $\endgroup$ – DukeZhou Jun 2 '18 at 3:24
3
$\begingroup$

If the example questions are not on-topic here, then I do not think any of them are unique, interesting and high enough quality to be worth migrating to Data Science stack exchange as they stand now.

As a long-term contributor to Data Science, my thoughts on the questions are:

  • Performance Evaluation Metrics used in Training, Validation and Testing is too vague and broad. Comments attempting to clarify with OP have not really resolved it. IMO, this might get answered on Data Science, but equally could be left abandoned as it is here, or closed as "Too broad" or "Unclear".

  • Forecasting and predict using matlab Artificial Neural Network looks like OP is trying to apply a regression model to a classification problem. However, there is nowhere near enough detail in the question to answer it well. Unless the OP was willing to get involved in clarifying how they are using the data set, this would likely get closed with "Unclear" on Data Science.

  • Deep NN architecture for predicting a matrix from two matrices could potentially be answered (I suggest a possible work around for OP in comments), and might be OK on Data Science if clarifying comments by OP were included. Data Science does get a lot of "I have a data set with this special trait, and I'm stuck about what to try" questions. Some are good, many are not clear, most just need the OP to go ahead and try stuff*. IMO, the question here is borderline - not quite enough information to make a good answer, but it can probably be answered. As such though, I'm not sure of the value of migrating it so long after it was asked. I think migrating a similar question in future would be well received.


* This is an ongoing issue on Data Science and I suggested we need to do something about it on Data Science meta a while ago. Maybe Data Science needs a help advice similar to Stack Overflow's excellent https://stackoverflow.com/help/mcve

$\endgroup$
  • $\begingroup$ The thing that confuses me is that some questions asked here have excellent answers on other stacks, recently i linked one of your answer from ds.se...so what abt those questions? $\endgroup$ – DuttaA Jun 2 '18 at 16:29
  • $\begingroup$ @DuttA: I think it is down to the detail in the questions. It is very hard to tell when to migrate, as you effectively need to be an expert on both sites to have a good chance of success. $\endgroup$ – Neil Slater Jun 2 '18 at 17:02
  • $\begingroup$ Thanks for weighing in on these three specifically. I've taken action to address, where possible. Definitely let me know in future when you think questions are best migrated, or if you think they have value on SE:AI. (Possibly we could post some of the AI questions in the chats for the related stacks, to try and attract more expert users contributing over here:) $\endgroup$ – DukeZhou Jun 4 '18 at 21:00

You must log in to answer this question.

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