@Glen_b It looks like you and I are racing to first in this week's reputation league. I think that's a personal best for me!

@Sycorax Hmm. I'm sure I can find something to upvote

@Glen_b Hah. That's not what I meant.

@Sycorax You plainly deserve to win a week. Half of my votes come from old threads, I think your votes are nearly all current.

@Glen_b Really the only answer that I think didn't get the recognition it deserves is the one about the definition of big data. stats.stackexchange.com/questions/173060/… but I think that lack of recognition dovetails nicely with my stance that almost no one knows what they're talking about re: big data

@Sycorax: Well, 13 votes isn't to be sniffed at!

Isn't perhaps someone who talks about "big data" - & not talking through his hat - really talking about the challenges posed by how memory (& time) requirements of data structures & algorithms grow with the size of the data set?

@Scortchi I have created a whole bunch of tag synonym suggestions (following the list in the relevant Meta discussion). In theory, anybody with >=5 score in the master tag can vote for or against, and a suggestion needs 4 votes to be implemented. In practice, it's not clear to me if anybody is going to notice and take part in voting, and for some suggestions there are less than 4 people with >=5 score...

Whoever reads this and has over 2.5k rep, please take a look at stats.stackexchange.com/tags/synonyms, maybe you can vote for some of the suggestions. The discussions are here: meta.stats.stackexchange.com/questions/1200

Some suggestions I could not create because I don't have sufficient score in the corresponding tags.

@gung I am sure you can vote for some.

I voted on a couple. I suggested [unbalanced-classes] ← [imbalanced], as well.

@nullgeppetto that would be a question for the main site. You have to define what "good estimation" is - the more closely you want to identify the parameter values (e.g. in a frequentist paradigm, the smaller the confidence region you want), the larger the sample you need. We can't tell you what "good" is for your purpose. However,

some relevant formulas may already be covered on site (e.g. a joint interval for the components of mu will be the interior of an ellipsoidal region based on a multivariate t-distribution; I expect that's on site already, the corresponding calculation for the variance-covariance matrix is probably discussed also; the joint calculation for mu with Sigma might not be); if you want Bayesian calculations you should indicate