Hot answers tagged

10

Although they happen to overlap a lot, I think they're still distinct. I would consider something like the Perceptron a neural network but not an example of a deep learning algorithm. Unless the tags are obviously redundant I wouldn't merge them.


9

Yeah, that tag is worthless - I've removed and blacklisted it. datascience was already blocked by default.


9

Yes, I think so, and this site is probably the best one for such questions. I tend to focus on operational aspects of analytics in work as a "data scientist" -- data cleaning, fusion, large-scale model building, serving, productionization. However I find the popular understanding of "data scientist" refers more to exploratory/investigative analytics -- ...


8

I think the tag is appropriate when scale is a non-trivial part of the problem. Not all data science involves huge amounts of data. While I agree that "big data" is an ill-defined term, I think it is relatively unambiguous here as a tag for questions where the challenge is partly data volume. Whether a given question really has anything to do with scale ...


7

This is a great question! Thanks for bringing it to meta. Question(s): I am going to summarize your question into two major points. Should we try to avoid creation of tags if another tag will work? How aggressively should we curate tags? TL;DR: Yes, if you are contributing content, then try to avoid creation of tags if another tag will work. You should ...


7

Visualization is often extremely helpful, although it can also (like any metaphorical technique) be used to coerce or obfuscate. So I would not consider it an integral part of data science myself. As to the site, if I can venture to suggest: Questions related to how visualizations and visual analytics can help one understand and convey topics in data ...


6

No - tags should not be a puzzle, they should categorize a puzzle. Tags inevitably become something that need to change over time, but, ..... .... I was going to say something derogatory about that tag taxonomy you describe on CSTheory, but, as I looked in to CSTheory to see how it works, I realized I don't understand how it works.... and, I think that is ...


5

I'd also say yes. What is visualization used for? During your own data exploration process, plotting is essential You have to communicate results to other scientists or peers from your experts group (goal: produce polished scientific graphs) You have to convince the non-expert stakeholders: the general public, your manager or supervisor, some ...


5

You are right that we need to have a pyspark tag. Users with 150 rep can already create tags. So, I have created one and would edit the questions accordingly. You can also help by identifying appropriate questions and sending in edit requests. Thanks for contributing.


4

Yes, when people use this tag, they almost always mean "recommender engine" here. Such questions should probably be tagged with recommender or recommender-engine instead. All the more so since recommendation is often used on SO to mean "asking for recommended resources", which is not the same thing nor a great type of question anyway.


4

I hadn't noticed specific-question; my first reaction was that it doesn't add much value to say "this is a question about something in particular" since presumably most questions are. But I get it. published-paper? published-something seems reasonable to denote that it's a question about known research, maybe.


4

For what it may be worth, I would have said time-series is the most commonly used term for this. I have never heard panel-data myself. longitudinal-data has come up but it seems far less used. And I would say I work in "data science". Hence my vote for favoring the time-series tag. I might even go so far as to call all of them synonymous on this site, but, ...


4

Thanks for bringing this concern to meta, and thanks for giving your time to make the site better. Attribution is needed: You state: I look on other Stack Exchange sites if they have a decent Tag description and I use that one as well by adding/removing parts that not make sense or are poorly written. And then ask: Is this acceptable? Based on this Meta ...


4

The rejection reasons sound like valid criticisms to me. In particular, since this tag seems more relevant for StackOverflow, a description for this site should ideally point out when this tag should be used here specifically. Generic questions about seaborn are more on-topic on StackOverflow. Hence, a description should point out when it would make sense to ...


4

Sounds reasonable. I went over the questions with the unbalanced-classes tag and agree they refer to the same thing. I'll go ahead and merge the two making class-imbalance the target tag.


3

here is the tag: https://datascience.stackexchange.com/questions/tagged/partial-least-squares


3

Sure, can you just suggest an edit? it can refer to predictive models in general. Yes, nothing specific to regression.


3

I am sorry this didn't draw a response. I feel neutral about the tag; I can imagine that questions only about datasets are better on OpenData, questions here might feasibly pertain in part to datasets. Your logic is sound though I could conceive of searching for "noise datasets" or something seeking questions about dealing with noise. I suppose that de ...


3

The tag is obviously redundant. But this is typical for all SE sites.


3

+1 to Sean Owen's answer. I also favor the time-series tag. panel-data is a common term in macroeconomics. The connotation is one of multiple time series measured over the same time, with interactions between the different series (possibly with lags). Example: a country's GDP, industrial production and joblessness, or multiple countries' GDPs. longitudinal-...


3

If you find a question that you believe could use this tag, and do not have the reputation to add it, then comment on the question and I will add it.


2

This is just my opinion, but I think of "text mining" when I think of search and retrieval mostly, but also document clustering and topic modeling. "Natural language processing" to me includes translation, sentiment analysis and entity recognition. I think they can reasonably considered different, but they overlap. You're welcome to try to list some key ...


2

Reproducing a similar question from the Cross Validated's meta forum. Creating tags just because there might be questions on the tag is unnecessary. However, if there are any questions which might use the tag, then creating one is very much advised.


2

Since no other proposals are forthcoming, here is one to consider: Keep all three of the tags mentioned above, but make panel-data and longitudinal-data synonyms. This will help users who search for one term by showing content tagged with either term. Wait until there are more questions under these tags to characterize the distinction between these and time-...


2

Before I start, I need to say that the referenced question, is a really great question. I hope it gets some good answers. And as to the inquiry about adding a new tag, thanks for bringing it to Meta. But... Let's ask ourselves why do we tag? If you follow that link, you will find (to my mind) that there are two primary benefits to tags. For me, the ...


2

Not possible However, you can add the tag to an existing question.


2

That sounds reasonable to me. I have implemented both of those mappings now.


2

I agree that if "energy" has any usage in this topic, it's when talking about ideas borrowed from other fields like the energy of a Hopfield network or something. It doesn't get used much. Is it wrong to tag a data science question about energy usage with "energy"? I don't think it's wrong, or at least not detracting. It's not like there's a common usage of ...


2

It is interesting you bring this up, I was thinking about this the other day. I think this is a good point. Perhaps it would make more sense to merge them? I have been adding the Orange3 tag to relevant questions recently thereby increasing it use - however I am not sure any question asked recently refer specifically to any earlier versions of the software. ...


1

In general, you can merge tags when you have enough total reputation and a positive answer score for the tag under consideration. A positive answer score is sufficient to vote on tag synonyms. hyperparameter seems to be more general than hyperparameter-tuning, for example, it could include questions about parameters of a Bayesian prior distribution.


Only top voted, non community-wiki answers of a minimum length are eligible