full2010.pdf - page 1516

1478
P54
„µ¦ž¦³¤µ–‡nµÂnªŠÎµ®¦´‡nµÁŒ¨¸É ¥ °Š„µ¦Â‹„‹ŠÁ¦ µ‡–·˜
On Confidence Intervals for Geometric Distribution
¤´¨¨·„µ šµœ» ·š›·Í1
Manlika Tanusit1
š‡´—¥n °
Šµœª·‹´¥œ¸Ê¤¸ª´˜™»ž¦³Š‡rÁ¡ºÉ°Áž¦¸¥Áš¸¥ª·›¸„µ¦ž¦³¤µ–‡nµÂnªŠÎµ®¦´ ‡nµÁŒ¨¸É¥ °Š„µ¦Â‹„‹Š
Á¦ µ‡–·˜ ×¥šÎµ„µ¦Áž¦¸¥Áš¸¥ 5 ª·›¸‡º° ª·›¸ž„˜· ª·›¸ ž„˜·ÂÁr ª·›¸ ‡°¦r ª· ›¸ ‡°¦rÁr ¨³ª·›¸°´˜¦µnªœ
£µª³œnµ‹³Áž}œ ץčo‡nµž¦³¤µ–´¤ž¦³· š›·Í‡ªµ¤ÁºÉ °¤´ÉœÂ¨³‡nµ‡ªµ¤„ªoµŠÁŒ¨¸É¥˜Éε»— °ŠnªŠ‡ªµ¤ÁºÉ°¤´ÉœÁž}œ
Á„–”rĜ„µ¦˜´—·œÄ‹ Ÿ¨„µ¦ª·‹´¥¡ªn µ ª·›¸š¸ ɤ¸‡ªµ¤Á®¤µ³¤Äœ„µ¦œÎµÅžÁ¨º°„čo‡º° ª·›¸°´˜¦µnªœ£µª³œnµ‹³Áž}œ
Á¡¦µ³Áž}œª· ›¸Á—¸¥ªš¸É¤¸‡nµž¦³¤µ–´¤ž¦³·š›·Í‡ªµ¤ÁºÉ°¤´ÉœÅ¤n ˜É優n µ‡nµ´¤ž¦³·š›·Í‡ªµ¤ÁºÉ°¤´É œš¸É„ε®œ—‹¹ŠœÎµÅž¼n
„µ¦‡Îµœª–®µ‡nµ‡ªµ¤„ªoµŠÁŒ¨¸É¥Å—oĜš»„Ç™µœ„µ¦–rš¸É šÎµ„µ¦ª·‹´¥
‡ÎµÎµ‡´ : „µ¦ž¦³¤µ–‡nµÂnªŠ „µ¦Â‹„‹ŠÁ¦ µ‡–·˜ ˜´ªž¦³¤µ–Ár
Abstract
The aim of this research was to compare interval estimation methods for the geometric distribution.
Normal method, Bayes Normal method, Score method, Bayes Score method and Likelihood ratio method were
considered. The approximate confidence coefficients and the lowest average width were used as the criteria to
selection. Results of the research were as follow. The likelihood ratio method was suitable for considered because
the method was given the approximate confidence coefficients not less than given confidence coefficients for any
condition were researchion.
Keywords: Interval estimation, Geometric distribution, Bayes estimator.
1 °µ‹µ¦¥r µ µª·µ™· ˜· ‡–³ª·š¥µ«µ˜¦r ¤®µª· š¥µ¨´¥Â¤nËo Á¸¥ŠÄ®¤n 50290.
Department of Statistics, Faculty of Science, Maejo University, Chiang Mai, 50290.
Corresponding author : Ú¦«´¡šr/Ú¦µ¦ 0 5387 3551-58 ˜n°105 E-mail:manlika@mju.ac.th