-2

I'm interested in hearing about what statistics problems other data teams are trying to tackle. For me, I have data on energy usage in the United States, and we are trying to figure out ways to forecast energy demand as well as figure out ways to balance heavy loads on the power grid. The statistical challenges are in short term and long term time series forecasting. Regime changes such as weekdays -> weekends further complicate the problem, as well as external factors like whether variables. Spatial correlation is also a nontrivial issue.

What type of data do you have, and what questions are you hoping to answer? What statistical challenges are there?

Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143
JCWong
  • 1,392
  • 1
  • 15
  • 29
  • Sorry Bogdanovist...I don't think the is/are "mistake" should warrant an immediate downvote of this question. Plus, isn't "is" the correct way to phrase this? The noun here is "your data team" which is singular. Imagine if you had a noun like "he", a singular noun, the question "What questions is he hoping to answer" makes sense, but "What questions are he hoping to answer" doesn't – JCWong Aug 31 '12 at 05:31
  • 1
    Just to clarify, I did not downvote this question. Incorrect grammar doesn't make a question bad and downvotes are for poor or problematic questions, not incorrect grammar. If you would like to discuss why your original sentence was incorrect, I'd suggest asking at the English Language and Usage stackexchange site. – Bogdanovist Aug 31 '12 at 05:44
  • take a look at e.g. http://www.kaggle.com/competitions – mlwida Aug 31 '12 at 07:35
  • 4
    (-1) Sorry, I am not able to see the value of this question, for me it is like asking "what do you currently code with R and what are the challenges you are facing". Yes, it can be extending one's view on the world, but it is not a good question according to SE-standards. [What is the definition of a list question?](http://meta.stackexchange.com/questions/124450/what-is-the-definition-of-a-list-question) – mlwida Aug 31 '12 at 07:41
  • I disagree - it would be both interesting and give practicing statisticians a chance to see what statistical areas they should spend some time educating them about. However, as it stands the question has too many tags and should also be community wiki. I'll put the later on my answer and edit to change the former. – Erik Aug 31 '12 at 08:31
  • 2
    "Is" is correct. The subject of the sentence is "team" not "questions". J.C. Wong is right. – Michael R. Chernick Aug 31 '12 at 10:32
  • Related or even a duplicate:[What is it that a statistician does?](http://stats.stackexchange.com/questions/5457/what-is-it-that-a-statistician-does?rq=1) – mlwida Aug 31 '12 at 12:12
  • 2
    @Michael: collective nouns such as "team", "Microsoft", "band", etc. are singular in American English, but not in British English. "Microsoft are working on this" is unobjectionable in BrE. See [Is “staff” plural?](http://english.stackexchange.com/questions/3288/), [Are collective nouns always plural, or are certain ones singular](http://english.stackexchange.com/questions/1338/), and the related questions linked from there. – RegDwight Aug 31 '12 at 12:50
  • 1
    @RegDwight Why make an issue about a noun being plural if it is at least singular to the American audience? – Michael R. Chernick Aug 31 '12 at 13:31
  • @steffen This is not a duplicate to what a statistician does. A data analysis team is composed of more than just statisticians and questions that a data analysis team wants to answer is more focussed than the work a statistician does is. – Michael R. Chernick Aug 31 '12 at 13:33
  • This is a question of about what are the typical topics your organization works on. I think it is broad but can readily be answered. – Michael R. Chernick Aug 31 '12 at 13:35
  • @Michael: That, in turn, you have to ask the editor. Or take it to your meta. I'm only here at all because the editor asked for clarification [via *our* meta](http://meta.english.stackexchange.com/questions/3049/can-we-get-some-help). – RegDwight Aug 31 '12 at 13:39

1 Answers1

1

I am working mostly with biological data, samples from humans and animals. Sometimes also cell cultures. It's in the -omics area so the ration of variables to observations is generally high (about 1:1, it's not genomics). So for visualization the usual techniques like PCA and PLS are very imporant.

There is also a large amount of standard linear modelling, where a problem is that we usually do not have the data to model the covariance structures and when we look at the response variables univariately we do not have the time to look at each model seperately, so we have to automate model selection and diagnostics as much as possible.

Finally, classification is also an imporant topic since diagnostic questions is often what brings customers to us.

Erik
  • 6,909
  • 20
  • 48