After a long discussion I had with the moderator oW_ about the community status in Data Science Site, I decided to gather some ideas and propose them here.

Note: My personal view is that a moderator should act as an Ambassador. Someone who will be responsible to promote and activate the community. Not just accepting/rejecting questions. However, I fully understand that it is just my personal view and behaviour that one should not demand from local staff members.

Let’s start with the ideas now

#1 Create useful content

Even if this is the first one, it’s not the most important. We already have enough questions and answer, but it’s something we can do in parallel. When we have knowledge of something useful, but there isn’t any related question already on the site, we can ask it, wait for 2-3 days and if there isn’t any decent answer, post the answer ourselves. This is something we did in Open Data SE.

What we achieve with this: Improve Google Ranking of DS SE and increase the traffic to our site.

#2 Reduce the number of unanswered questions

If we all try to answer 1-2 questions from the unanswered list every week, we will manage to keep it low.

What we achieve with this: It’s not about the number, but make users feel that there will always be an answer for them.

#3 Guides and Tutorials

We can have community-wiki guides about specific parts of the daily routine in Data Science SE. It’s doesn’t have to be long, endless documentation. It can be in a format of QnA. For example, how should I judge if I should vote to close a question? Create a step by step guide of what a user must examine. Giving links to the Help section, adding screenshots and comments.

What we achieve with this: We help users to contribute to the site. We don’t let them find the right way themselves.

#4 Special Days - Datathons

We can schedule it that once every x months, we create a community event with a specific target. For example, we will try to answer all unanswered questions with that specific Tag. We promote it inside DS Site and outside in social media. We have an open call for anyone to participate.

What we achieve with this: Participation. Any user wants to feel that is part of a bigger community. Let them feel it.

#5 Special Days - Self-promotions

I think I saw it in the past in another Meta site. There is one special day every month where a user can promote a project related to Data Science in a meta question that started by a mod. It shouldn’t be a link promotion, but a short-medium post about how this project is related to DS. It’s not necessary that users will promote their own projects. They might do any interesting project they found.

What we achieve with this: We create an extra bond between users and DS Site. Also, they have an extra reason to visit Meta.

#6 Chat rooms

We don’t use chat rooms at all. We don’t have to do it when there is a long comment discussion. We could have different chart rooms for different generic subjects (about Data Science) and promote the discussion there as well.

What we achieve with this: We keep users more active during their session here

#7 Site promotion

I put this one last since it’s not something that anyone might want to do. Promotion of the Data Science SE in other places. Last February, I did promote the website in two big communities outside of Stack Exchange. I made it clear that the link is not an affiliate, I don’t have something to win from this and ask for specific actions. Here is a screenshot of one of the posts I did.

enter image description here

I believe that the results of those two promotions were big enough. If we check the number of questions/answers during that period we can see a change in the trend.

enter image description here

Those were some of the ideas I have to improve this community that we all love. Please use the answers to either add feedback on those specific ideas or suggest another one that you like. It might be better if we decide to coordinate one of those, to do it through a chat room since it might need a longer discussion.

  • 1
    $\begingroup$ #0 merge with CV ? $\endgroup$ Aug 12, 2019 at 16:18
  • 2
    $\begingroup$ @were_cat I disagree, data science is very broader than what we have in CV. $\endgroup$ Aug 13, 2019 at 16:34
  • 1
    $\begingroup$ @Vaalizaadeh That's true. There are questions that cannot be answered in CV and vice versa. $\endgroup$
    – Tasos
    Aug 13, 2019 at 17:26
  • 1
    $\begingroup$ @Tasos I believe the reason that explains why the members do not interact is due to the fact that our users do not try to upvote answers which are valuable. You can easily see questions that have more than 100k views, but the users do not upvote them. The counter is increased for registered users who can upvote, but they simply do not :) I really do not know, but I do believe that we have a better community than SO. The reason is that here, our friends do not try to humiliate the questions while this is a common approach for SO users. I really do not see any fun to humiliate someone who $\endgroup$ Aug 13, 2019 at 18:41
  • $\begingroup$ is trying to find his answers and improve his knowledge, but unfortunately, we see that among SO users. Nevertheless, I'm very proud of our community. Our field is young and will grow, as our users :) $\endgroup$ Aug 13, 2019 at 18:42
  • 1
    $\begingroup$ If there is some meaningfull distinctions between statistics and data science in general, I feel like those differences aren't enough to justify the creation and maintenance of another SE. I honestly can't see anyone that would only be interested in DS but not CV. In my opinion this lack of specific demand is the original sin that condemn DS to low volume and poor user engagement. Maybe the CV SE can be broadened a bit, the DS questions handeled here with a tag... (I honestly can't see why the programming questions are not redirected towards SO). $\endgroup$ Aug 14, 2019 at 16:25
  • 2
    $\begingroup$ Look at the comments on the reddit post mentionned above : threes comments asking about the articulation with an already existing SE, one saying that the SE format is not adapted to Data Science... $\endgroup$ Aug 14, 2019 at 16:30
  • 1
    $\begingroup$ Would it be possible to analyse frequently asked questions to see what would be the most useful community wikis to make. If half of new users questions are "what algorithm should i use for X" then we should make a wiki listing resources for that. $\endgroup$
    – Tasty213
    Aug 22, 2019 at 10:38
  • $\begingroup$ @Tasty213 that's a nice idea. $\endgroup$
    – Tasos
    Aug 22, 2019 at 12:47
  • 1
    $\begingroup$ I agree with @Media I get the same vibe, I also find many answers that are valuable but are neither accepted nor upvoted. $\endgroup$
    – 20-roso
    Aug 25, 2020 at 19:51

2 Answers 2


Ideas 8: Reduce the overlap with http://stats.stackexchange.com. Currently, many questions posted on https://datascience.stackexchange.com are also on topic on http://stats.stackexchange.com, and duplicating information between Stack Exchange websites isn't great for convincing people to contribute (not to mention the existence of https://ai.stackexchange.com).

  • 1
    $\begingroup$ That's true. It's the first question I got from someone when I discuss about the DS StackExchange. As a side note, I would like to see an improvement on closing vote where you can choose another SE site. At the moment, only meta is available, while both stats and stackoverflow should be there. $\endgroup$
    – Tasos
    Aug 26, 2019 at 11:14

For me, an important component is the community and the culture that goes with it. Data Science being quite multidisciplinary brings a few people and fields from different backgrounds. Also, Data Science is in its infancy compared to statistics (Cross Validated) and other domains.

These I think answer the question of why we have this influx of new questions and their answers lagging. Is not consistent. Questions come from every person that wants to do some sort of data analysis and people who can answer those questions may not necessarily be in the Data Science SE (maybe browsing mostly Cross Validated for instance).

I agree with the other answer, in addition, I think it would be a good idea to kind of purge the SE Data Science, maybe even rename it. Purpose it only for some Data Science aspects, like machine learning and its deployment, computer vision, NLP, etc, no questions on how to handle Python or R and how does linear regression work or what kind of regression is best.

I believe in increasing the number of moderators by rewarding them on cleaning questions that do not belong here or are of poor quality. Also, upvote answers which are useful, as it is mentioned in the comments above that another cause of our community not being active is not rewarding valuable answers.


You must log in to answer this question.

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