µ¦¦³»
¤ª·
µµ¦Â¨³Á°¨µª·
´
¥ ¤®µª·
¥µ¨´
¥´
¬·
¦´Ê
¸É
19 ¦³Î
µe
2552
µ¦Á¦¸
¥Á¸
¥ª·
¸
µ¦Â¨
o
°¤¼
¨¸É
¤¸
µ¦ÂÂŤn
·
®¨µ¥´
ªÂ¦
Įo
¤¸
µ¦¨¼n
Á
o
µ¼n
µ¦Â·
®¨µ¥´
ªÂ¦
Comparison of Data Transformation Methods of
Non-Multinormal Distribution to Multinormal Distribution
»
¡¦ ´Ê
µ£µr
1*
Jatuporn Tangkanjanapas
1*
´
¥n
°
µ¦ª·
´
¥Ä¦´Ê
¸Ê
¤¸
ª´
»
¦³r
Á¡ºÉ
°«¹
¬µÂ¨³Á¦¸
¥Á¸
¥¦¼
µ¦Â¨
o
°¤¼
¨¸É
µ¦ÂÂŤn
·
®¨µ¥´
ªÂ¦Ä®o
¤¸
µ¦¨¼n
Á
o
µ¼n
µ¦Â·
®¨µ¥´
ªÂ¦ Ã¥¡·
µ¦µµ¦Â¨
o
°¤¼
¨´Ê
®¤ 3 ª·
¸
º
° µ¦Â¨
o
°¤¼
¨Â â¦r
¡µªÁª°¦r
(folded power transformation) µ¦Â¨
o
°¤¼
¨Â Guerrero-Johnson (Guerrero-
Johnson transformation) ¨³µ¦Â¨
o
°¤¼
¨Â Á¨ °¼
¨µ (scaled angular transformation) Ã¥¡·
µ¦µµ
Á°¦r
Ár
¼
»
°µ¦¥°¤¦´
µ¦Â¨
o
°¤¼
¨ªn
µÅo
¤¸
µ¦Â·
®¨µ¥´
ªÂ¦µµ¦°
¤¤·
µ (
o
°¤¼
¨¤¸
µ¦Â·
®¨µ¥´
ªÂ¦) Á¡ºÉ
°Ä®o
¦µªn
µ¦¼
µ¦Â¨
o
°¤¼
¨ÂÄÁ®¤µ³¤
´
o
°¤¼
¨»
´Ê
Ç ¤µ¸É
»
£µ¥Äo
µµ¦r
¸É
Î
µ® º
° ¦³µ¦¤¸
µ¦ÂÂÂÁ¨
¸Ê
Î
µ¨´
µ¦ÂÂ
ÂŪ¼
¨¨r
µ¦Â¤¤µ ¨³µ¦ÂÂÂÁµ ¹É
Î
µ®Ä®o
¤¸
¨´
¬³Ãn
Áo
n
µÇ´
µ
´
ª°¥n
µ º
° 20 , 30 , 50 ¨³ 100 ¦³´
´
¥Î
µ´
°µ¦°Á¸
¥ªµ¤¨¤¨º
0.05
ĵ¦ª·
´
¥¦´Ê
¸Ê
Î
µ¨°µµ¦r
µ¦¨°o
ª¥Á·
¤°·
µ¦r
è ¹É
Î
µµ¦¨°ÊÎ
µÄÂn
¨³¦°
ªn
µ¨n
µ¦³®ªn
µ
X
~
Ħ°¸É
i+1 ´
¦°¸É
i ¤¸
n
µo
°¥¤µº
°ªn
µÅ¤n
Ân
µ´
Ã¥¤¸
Ár
ªn
µ
0001 .0 ~ ~
1
d
i
i
X X
(Ã¥¸É
i = 1,2,…,k ¨³ i º
°Î
µª¦°Äµ¦¨°) ¹
®¥»
µ¦¨°
µ¦ª·
´
¥¦´Ê
¸Ê
¦»
Åo
ªn
µ Î
µ®¦´
»
µ¦ÂÂÄ»
¦³´
°ªµ¤Áo
n
ªÄ®n
ª·
¸
µ¦Â¨Â â¦r
¡µªÁª°¦r
¨³ª·
¸
µ¦Â¨Â Guerrero-Johnson Ä®o
Á°¦r
Ár
µ¦¥°¤¦´
¤¤·
µ¼
¸É
»
Î
µÎ
µ´
: µ¦Â¨
o
°¤¼
¨ ; µ¦Â·
®¨µ¥´
ªÂ¦ ; µ¦ÂÂŪ¼
¨¨r
Abstract
The objective of this research was to investigate and compare the data transformation forms which can
transform data to multinormal distribution. The data transformation forms were folder power transformation,
Guerrero-Johnson transformation and scaled angular Transformation. The highest percentage of accepted
multinormal distribution was used in evaluating the capability of the most appropriate data transformation forms
1
·
·
µ
µª·
µ·
·
®¨´
¼
¦¦·
µ·
·
«µ¦¤®µ´
·
³¡µ·
¥«µ¦r
¨³µ¦´
¸
»
¯µ¨¦r
¤®µª·
¥µ¨´
¥
*
Corresponding author : æ«´
¡r
/浦 081-6442979