Sometimes I feel I have some questions that people that just learned from book will not understand what I was asking, but some really good people feel it is no need to explain. How can I ask those questions?

One example is this question, Why there are two different logistic loss formulation / notations? I think I struggled for a long time on deriving two notations are the same in different optimization setting.

But a person just learn logistic regression form few classical statistics books may not seen another notation, and experienced people that writing statistics or machine learning software may think this is trivial just a notation problem.

Am I right such questions exist? Should I ask questions like this? Any other suggestions?

  • 1
    $\begingroup$ In this case, as you now know, it comes down to a trivial difference in how binary results are coded. Finding such things out (especially when you're reading textbooks) is analogous to the expectation on Stack Overflow that you carry out some basic debugging (which means consulting your sources for their definitions and formulas), reduce your problem to a minimal reproducible example, and search for similar situations on the site (as one user commented in your question). If, after going through that research you still have a question, you will likely be able to formulate it clearly. $\endgroup$
    – whuber Mod
    Aug 15, 2016 at 17:34
  • $\begingroup$ @whuber so your suggestion would be holding immature questions? Sometimes I want to try to ask. since may be there are other people suffer from the same problem can give me answer. $\endgroup$
    – Haitao Du
    Aug 15, 2016 at 18:01
  • 4
    $\begingroup$ Ultimately I would like to suggest following all relevant advice for formulating good questions. Whether they are "mature" or not, advanced or not, difficult or not, makes little difference. $\endgroup$
    – whuber Mod
    Aug 15, 2016 at 19:54
  • 6
    $\begingroup$ I would say, 'do your best', & go ahead and ask. It's OK (or should be) if it turns out that the question was based on a misunderstanding of some type. We should be sufficiently forgiving of people trying to ask questions that they might be mixed up. $\endgroup$ Aug 15, 2016 at 21:17
  • 4
    $\begingroup$ hxd I take whuber's advice not to be to "hold" such questions but simply to do the appropriate kinds of research first in order to be able to formulate the ideas properly. This is simply part of the necessary process to ask suitable questions. This process will often answer the question along the way (since you gain some of the understanding necessary to answer a question at the same time you gain the necessary understanding to formulate it). Many questions I think to ask I answer in this fashion (but sometimes I still ask them anyway, often because it's useful to have a good answer on site). $\endgroup$
    – Glen_b
    Aug 15, 2016 at 22:38
  • 3
    $\begingroup$ Asking good questions is a nontrivial exercise - I find it often takes more mental effort to formulate a good question than to construct a good answer; it's a skill that takes thought and practice. $\endgroup$
    – Glen_b
    Aug 15, 2016 at 22:42


You must log in to answer this question.

Browse other questions tagged .