Talk:Imputation (statistics)

Untitled

Mean Imputation is the best Imputation. —Preceding unsigned comment added by 220.225.85.241 (talk • contribs)

An interesting, but pretty debatable, proposition, unsigned. Randomactsofaffect (talk) 15:18, 15 May 2010 (UTC)[reply]

Wiki Education Foundation-supported course assignment

This article is or was the subject of a Wiki Education Foundation-supported course assignment. Further details are available on the course page. Student editor(s): Cki17. Peer reviewers: Cki17.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 00:21, 17 January 2022 (UTC)[reply]

Intent to remove this misinformation

This article is about science fiction. It may be an attempt by Medscape or by someone related to Medscape, to create a false impression of truth. Medscape posted an interview dated May 5, 2011 that may be an April Fool's joke (a late one). The definition given for the term describes a a method that cannot be used to produce valid statistical results. The interviewee refers to the alleged existence of peer-reviewed work in either math or stats, that supports what he wrote. What isn't clear -- not even in the discloser -- is who wrote the content. Is the interview merely a performed script? It also fails to make clear that the interviewee has no medical training. The transcript refers to him as "Doctor" but doesn't explain in a place obvious to the reader, he is not a physician. This fact is obfuscated because the interviewer is an MD.

If you know of a WP rule that allows speedy deletion for articles like this one please content me via WP email. — Preceding unsigned comment added by Kernel.package (talk • contribs) 20:44, 29 May 2011 (UTC)[reply]

Wikipedia articles are not the place to publish your opinions, however strongly held. I have reverted them. You are welcome to open an deletion discussion for this article if you like, but I think you will be laughed out of court. The interview you object to ([1]) is a very brief piece that seems aimed at educating Medscape readers about imputation methods. It is not linked or cited by this article, not should it be IMO. There are much better popularisations of the topic available if one is needed. --Avenue (talk) 04:54, 30 May 2011 (UTC)[reply]

Critique concerning citations and lack of information

This wikipedia article is pretty barren given the recent amount of publications that have been released concerning imputation, particularly multiple imputation. Only a few facts are cited, so more citations must certainly be added. While the current group of sources seem to come from reliable sources, the article should draw upon more sources. In particular, the article only seems to use 4 out of 8 sources in its list of citations. While the article does not seem to be biased at all, the bias/disadvantages of single imputation are heavily understated. In addition the additional computation needed for multiple imputation needs to be mentioned as well. It is only tangentially touched upon with the sentence about machine learning. The multiple imputation section is relatively small and should be expanded. Cki17 (talk) 00:26, 28 September 2016 (UTC)[reply]

Dark matter as a result of imputation

Mention also the imputative dark matter theory. (gravitational curvature compresses the probabilities of free motion making some paths more likely, so the unlikely paths at a huge order of magnitude contribute to dark matter. I don't support it, but mention it.) — Preceding unsigned comment added by 2a02:587:4110:5800:159a:b911:fc8c:dd4c (talk) 09:49, 24 April 2017‎ (UTC)[reply]

The previous comment has nothing to do with missing elements in data sets, which is what this topic is about. ("Imputation" has more than one meaning, but the above comment should be placed in the WP topic on "Dark Matter". Imputation methods as discussed here attempt to address a very practical problem encountered, for example, with medical data, where different clinical parameters are sampled at different frequencies because of considerations such as expense or risk to the patient. (For example, you can readily measure body weight and blood pressure, but Magnetic Resonance imaging is expensive, and repeated liver biopsies are painful and hazardous to the patient. Prakash Nadkarni (talk) Aug 19th 2017.

kNN imputation

I've only learned of the term imputation last Friday, so not going to add anything to this article. To me, k-nearest neighbours seems like a natural way of imputing data. I don't know how it compares to the other methods in terms of prominence, but some scientific articles have been written about it:

Could anybody more knowledgeable about this add this or tell me why it shouldn't be added? Femke Nijsse (talk) 10:05, 21 May 2019 (UTC)[reply]

There's no reason not to add it, just nobody's gotten around to it yet (or, be bold and add it yourself!).Citing (talk) 00:47, 24 May 2019 (UTC)[reply]