Minimum distance estimation
From Wikipedia, the free encyclopedia
Minimum distance estimation (MDE) is a statistical method for fitting a mathematical model to data, usually the empirical distribution.
Contents |
[edit] Definition
Let be an independent and identically distributed (iid) random sample from a population with distribution and .
Let be the empirical distribution function based on the sample.
Let be an estimator for . Then is an estimator for .
Let be a functional returning some measure of "distance" between the two arguments. The functional is also called the criterion function.
If there exists a such that , . is called the minimum distance estimate of .
[edit] Goodness of fit statistics based on minimum distance estimation
The theory of minimum distance estimation underlies various statistical goodness of fit tests. The difference between the various tests is usually the selected "distance" estimator, or criterion. Below are some examples of statistical tests that rely on minimum distance estimation.
[edit] Chi-square test
The chi-square test uses as its criterion the squared difference between the empirical distribution and the estimate, weighted by the estimate for that group.
[edit] Cramér-von-Mises criterion
The Cramér-von-Mises criterion uses the squared difference between the empirical distribution and the estimate.
[edit] Kolmogorov-Smirnov test
The Kolmogorov-Smirnov test uses the supremum of the absolute difference between the empirical distribution and the estimate.
[edit] Anderson-Darling test
The Anderson-Darling test is similar to Kolmogorov-Smirnov, but instead of using the point of maximal difference between the empirical distribution and the estimate, it uses a smooth weighting based on integrating the difference over the entire interval and then weighting by the reciprocal of the variance.
[edit] See also
[edit] References
- Blyth, Colin R. (June 1970). "On the Inference and Decision Models of Statistics" (PDF). The Annals of Mathematical Statistics (Institute of Mathematical Statistics) 41 (3): 1034–1058. doi: . ISSN 0020-3157. http://projecteuclid.org/DPubS/Repository/1.0/Disseminate?view=body&id=pdf_1&handle=euclid.aoms/1177696980. Retrieved on 24 September 2008.
- Drossos, Constantine A.; Andreas N. Philippou (December 1980). "A Note on Minimum Distance Estimates" (PDF). Annals of the Institute of Statistical Mathematics (Institute of Statistical Mathematics) 32 (1): 121–123. doi: . ISSN 0020-3157. http://www.ism.ac.jp/editsec/aism/pdf/032_1_0121.pdf. Retrieved on 24 September 2008.
- Wolfowitz, J. (March 1957). "The Minimum Distance Method" (PDF). The Annals of Mathematical Statistics (Institute of Mathematical Statistics) 28 (1): 75–88. doi: . ISSN 0020-3157. http://projecteuclid.org/DPubS/Repository/1.0/Disseminate?view=body&id=pdf_1&handle=euclid.aoms/1177707038. Retrieved on 24 September 2008.