Talk:Density estimation

Hello. I've pasted in an example which used to be at Illustration of density estimation until that page was deleted by people who apparently mistook it for original research (see Articles for deletion/Illustration of density estimation). Oh well. In any event, the example is probably happier here. It may be a little too detailed -- maybe the raw data and the script should be moved elsewhere (e.g., to one of the image description pages). I think it's pretty important to keep that info somewhere within reach, though. Best, Wile E. Heresiarch 02:19, 18 June 2006 (UTC)

I agree with Wile that the data themselves aren't necessary here. I think there is enough info on the page to let people find the data. R is free so there is little impediment. I'll remove the data. Steev 03:43, 14 November 2006 (UTC)

I have added vector quantization to the list of techniques that can be used for density estimation Pog 22:05, 1 August 2007 (UTC)

Example very helpful. thanks. —Preceding unsigned comment added by 194.254.90.40 (talk) 15:17, 13 April 2008 (UTC)

Multiple issues: WP:NOTHOWTO and WP:INCITE
I've just tagged the page for multiple issues. The phrase "quoted verbatim from the data set description" raises a red flag: where exactly does the quotation come from? I found the same text at, so I'm guessing that the source is one of the two references Smith et al 1988 or Ripley 1996 mentioned on that page. It would be good to have an inline citation here to make it clear. And the rest of the page reads too much like a textbook or how-to manual rather than an encylopedia article; compare Kernel density estimation for an example of how to write in encyclopedic style about a similar topic. Jowa fan (talk) 06:05, 7 August 2012 (UTC)

Unsatisfactory
Not clear enough for the non-expert. For example the introduction "In probability and statistics, density estimation is the construction of an estimate, based on observed data, of an unobservable underlying probability density function. The unobservable density function is thought of as the density according to which a large population is distributed; the data are usually thought of as a random sample from that population." is ambiguous. Is density estimation merely an attempt to infer the actual distribution that would be followed by a very large data set, or is it an attempt to identify the "underlying" process that is generating the observations? Or are these the same thing, because the only truly accurate description requires identification of the generating process. I find the next section, the Pima Indian example, cumbersome, requiring programming knowledge, and unilluminating. It would be much better to use a simple mathematically well-defined example, with ample graphs, along the lines of the helpful fig 1 (which however isn't referred to or explained in the Introduction).Paulhummerman (talk) 13:49, 27 September 2018 (UTC)

Prompt
Density estimation is also used in generative models like PixelCNN and PixelSnail. (Neural Networks that can generate images.) I might come back to write about it myself, but due to my bad experiences with contributing to Wikipedia I leave it to you degenerates to take care of it for now.

https://arxiv.org/abs/1712.09763

https://arxiv.org/abs/1606.05328