Monte Carlo goodness-of-fit tests for discrete data [electronic resource] / Asya Rabinovich Takken

Takken, Asya Rabinovich
Bib ID
vtls000568521
稽核項
151 p.
電子版
附註項
數位化論文典藏聯盟
預約人數:0
全部評等: 0
沒有紀錄。
 
 
 
02269nam a2200325 a 4500
001
 
 
vtls000568521
003
 
 
VRT
005
 
 
20100107095100.0
006
 
 
m        d
008
 
 
030114s2000    miu      m    000 0 eng d
035
$a (UnM)AAI9961968
039
9
$a 201001070951 $b sujenlee $y 200301141650 $z sujenlee
040
$a UnM $b eng $d TKU
100
1
$a Takken, Asya Rabinovich, $e author
245
1
0
$a Monte Carlo goodness-of-fit tests for discrete data $h [electronic resource] / $c Asya Rabinovich Takken
300
$a 151 p.
500
$a Source: Dissertation Abstracts International, Volume: 61-02, Section: B, page: 0923.
500
$a Adviser:  Persi Diaconis.
502
$a Thesis (Ph.D.)--Stanford University, 2000.
520
$a This dissertation derives and tests Markov chain Monte Carlo algorithms for sampling from exact conditional distributions given sufficient statistics, in some discrete exponential families. These sampling distributions are used for hypothesis testing and confidence interval estimation when the data set is too small for asymptotic results to be reliable, but too large for exact testing. To sample from the conditional distribution, a random walk is constructed with state space consisting of all the data sets with the given sufficient statistics, and with a stationary distribution which is the desired conditional distribution. Most of the results in this work are concerned with finding a set of moves defining such a random walk, for a variety of models.
520
$a Models for which a set of moves is derived include poisson regression and logistic regression with integer or equally spaced or categorical covariate(s), common odds ratio in 2 x <italic>J</italic> x <italic>K</italic> contingency tables, and decomposable log-linear models.
591
$a 數位化論文典藏聯盟 $b PQDT $c 中山大學(2001~2002)
650
0
$a Statistics.
650
0
$a Mathematics.
700
1
0
$a Diaconis, Persi, $e advisor
710
2
0
$a Stanford University.
773
0
$t Dissertation Abstracts International $g 61-02B.
856
4
0
$u http://info.lib.tku.edu.tw/ebook/redirect.asp?bibid=568521
999
$a VIRTUA40         m
999
$a VTLSSORT0010*0060*0080*0350*0390*0400*1000*2450*3000*5000*5001*5020*5200*5201*6500*6501*7000*7100*7730*8560*9992
沒有評論
標題
摘要
This dissertation derives and tests Markov chain Monte Carlo algorithms for sampling from exact conditional distributions given sufficient statistics, in some discrete exponential families. These sampling distributions are used for hypothesis testing and confidence interval estimation when the data set is too small for asymptotic results to be reliable, but too large for exact testing. To sample from the conditional distribution, a random walk is constructed with state space consisting of all the data sets with the given sufficient statistics, and with a stationary distribution which is the desired conditional distribution. Most of the results in this work are concerned with finding a set of moves defining such a random walk, for a variety of models.
Models for which a set of moves is derived include poisson regression and logistic regression with integer or equally spaced or categorical covariate(s), common odds ratio in 2 x <italic>J</italic> x <italic>K</italic> contingency tables, and decomposable log-linear models.
附註
Source: Dissertation Abstracts International, Volume: 61-02, Section: B, page: 0923.
Adviser: Persi Diaconis.
Thesis (Ph.D.)--Stanford University, 2000.
數位化論文典藏聯盟
合著者