Talk:Precision (statistics)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Statistics (Rated Stub-class, High-importance)
WikiProject icon

This article is within the scope of the WikiProject Statistics, a collaborative effort to improve the coverage of statistics on Wikipedia. If you would like to participate, please visit the project page or join the discussion.

Stub-Class article Stub  This article has been rated as Stub-Class on the quality scale.
 High  This article has been rated as High-importance on the importance scale.
 

Bayes inference[edit]

PRECISION is used in BUGS (Bayesian inference Using Gibbs Sampler) software[edit]

PRECISION as used in Bayesian Statistical Packages The widely used BUGS family of languages (WinBUGS, OpenBUGS and JAGS) apparently uses "precision" (defined as the inverse of the variance) in specifying distributions. I am writing an article on the new Bayes Statistical package Stan (developed 2012, with version 1.2, 6, March 2013). Stan use the more typical Standard Deviation.

Here is a reference (page 247) in "Stan User's Manual and Reference Guide, Version 1.2, 6 March 2012 (Creative Commons)"

"In Stan, the second argument to the "normal" function is the standard deviation (i.e., the scale), not the variance (as in Bayesian Data Analysis) and not the inverse-variance (i.e., precision) (as in BUGS). Thus a normal with mean 1 and standard deviation 2 is normal(1,2), not normal(1,4) or normal(1,0.25). Similarly, the second argument to the "multivariate normal" function is the covariance matrix and not the inverse covariance matrix (i.e., the precision matrix) (as in BUGS). The same is true for the "multivariate student" distribution." [1]

If you type "BUGS reciprocal variance precision" into a search engine (I used Google) you will get several hits with BUGS using the word "Precision" to represent inverse variance.

Here is a lecture I found which also references the statistical language R:

"BUGS code

   BUGS code looks just like R code but with important differences.
   ...
   BUGS uses R probability functions but with a twist. For instance the dnorm function is parameterized in terms of precision (reciprocal variance) rather than the standard deviation."

[2]

Opinion: PRECISION article needed because term in in common use in BUGS (Bayesian inference Using Gibbs Sampler) software[edit]

So, IMHO, it would be helpful to have a separate article on "Precision" used in the sense of "inverse variance" that explains this specific use of the term and what relation (if any?) it has to other uses of the term "Precision" (as in rounding of calculations or precision of estimate). Jim.Callahan,Orlando (talk) 14:18, 11 March 2013 (UTC)

This is specifically the topic of this article. THe other meaning is covered by the second sentence "This is in addition to its more general meaning in the contexts of accuracy and precision and of precision and recall. " ... which presumably points reraders to those artcles. 81.98.35.149 (talk) 22:28, 11 March 2013 (UTC)

AGREE -- I was responding to a notice that this article might be deleted. I wanted to explain rationale why I thought it was needed. Jim.Callahan,Orlando (talk) 06:32, 12 March 2013 (UTC)

Weighting[edit]

Speculation/Question: Does the definition of PRECISION have something to do with Inverse-variance weighting?[edit]

SPECULATION -- Defining Precision as the reciprocal of the variance, might have something to do with Inverse-variance weighting? Jim.Callahan,Orlando (talk) 06:32, 12 March 2013 (UTC)

see also: Weighted mean#Dealing with variance Jim.Callahan,Orlando (talk) 12:24, 12 March 2013 (UTC)

Speculation/Question: Does the definition of PRECISION have something to do with Weighted least squares?[edit]

"The Gauss–Markov theorem shows that, when this is so, beta-hat is a best linear unbiased estimator (BLUE). If, however, the measurements are uncorrelated but have different uncertainties, a modified approach might be adopted. Aitken showed that when a weighted sum of squared residuals is minimized, beta-hat is BLUE if each weight is equal to the reciprocal of the variance of the measurement."

  • * *

"If the errors are correlated, the resulting estimator is BLUE if the weight matrix is equal to the inverse of the variance-covariance matrix of the observations." Jim.Callahan,Orlando (talk) 12:24, 12 March 2013 (UTC)

The above two points may be true algebraically, but to be included in this article you would need to find a source (or [referably sources) in which the weighting is done with things that are actually identified as the "precision", or as being equivalent to weighting using precisions. Recall that Wikipedia is based on reporting reputable sources, not on including original research ... WP:OR. 81.98.35.149 (talk) 15:36, 12 March 2013 (UTC)