Supposing I have data which I know is normally distributed, but because the recording process is right censored, how do I estimate the parameters of the distribution?
Asked
Active
Viewed 861 times
2
-
2One solution (out of many possible) is given, with working code, at http://stats.stackexchange.com/a/49456. – whuber Aug 14 '15 at 04:53
-
My answer was wrong, so I deleted it. Probably the formula's you are looking for are on page 20 of http://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf – Aug 14 '15 at 09:44
-
@fcoppens Your (deleted) answer and your comment confuse censoring with truncation. It is important to distinguish them and treat them differently. – whuber Aug 14 '15 at 13:42
-
Thanks @whuber! Do you happen to have pointers to any of the other many possible methods? Or at least search terms I should use? – 2daaa Aug 14 '15 at 13:57
-
Censoring is a big deal in some specialized (but popular) fields, which are busy rediscovering each other's methods. One set of methods comes from *survival analysis.* Another is developed in *chemometrics* (which is primarily concerned with left-censoring, but everything applies to right-censoring too). See [Dennis Helsel's book](http://www.wiley.com/WileyCDA/WileyTitle/productCd-EHEP002278.html), for instance. – whuber Aug 14 '15 at 14:18
-
2It looks like the answer @kjetilbhalvorsen cited does answer the question, but the title of the question is so different that it looks like a completely different question – Peter Flom May 02 '17 at 11:25