I have a question that requires knowledge of machine learning and a little knowledge of the Earth's ionosphere, specifically involving Total Electron Content (TEC). It would be about the minimum amount of TEC map data (period in years) that would be enough for me to obtain a reasonable data distribution so that my time series of this data is reasonably forecastable via neural network (considering the behavior of the solar cycle, for example, which is related to the TEC). What would be the best site to post this question?
4
-
6I'm not knowledgeable in either, but by the sound of it, it looks like it's more related to ML, so maybe Data Science is more appropriate in this regard. However, I have no idea if your specific question is a good fit on their site, I would read their How to Ask page first. I'd focus my question on how to identify how much data is enough to build your model, the TEC subject is not relevant on their site. That's the best I can suggest, hopefully for you someone with more knowledge in this matter will be able to provide better help.– LafCommented Apr 8 at 21:32
-
1@Laf ok, thank you very much for your suggestion, maybe it makes sense not to focus on the specific type of my data, I'll consider posting on Data Science's Stack!– MarcoCommented Apr 8 at 21:43
-
4There is an Earth Science SE that might be related to your topic– user13267Commented Apr 9 at 3:37
-
2CrossValidated SE is a Q&A community for people interested in machine learning (and statistics, etc.). You might look there as well for related existing posts on training parameters.– hardmathCommented Apr 10 at 11:29
Add a comment
|
1 Answer
Like hardmath points out in the comments, I think CrossValidated is the better/best option:
On the help/on-topic page of CrossValidated, it mentions how the site
is for statisticians, data miners, and anyone else doing data analysis or interested in it as a discipline. If you have a question about [..] designing experiments, collecting data, machine learning [..] then you're in the right place.
It seems you are interested in the minimum sample size of your data. You can look for similar questions using the machine-learning and sample-size tags (e.g. their combination here).
Data Science has the following comments on their help/on-topic page:
If you think a question is equally appropriate on multiple sites, ask on the site with the most users (usually Stack Overflow or Cross Validated). That way you have the best chance to get good and quick answers and site contents will stay more coherent. If it is not accepted there, it can be migrated to the correct site. Don't post your questions on more than one site.
Questions are most appropriate here if they are concerned with putting statistical concepts into practice, focus on implementation and (business) processes. Compared with statistics, data science is concerned with implementing whole analytical systems that can ingest (mainly large and diverse) data sets and estimate quantities of interest by incorporating advances from multiple fields.
But the answer you seem to be looking for might depend on the nature of your data, and you might want to break the problem up in two separate questions, where one asks after sample size, and the other to get a clearer idea of how your specific data set should be structured.
