1 .1. C(x) : refA, p.25
3 . C(x) : refA, p.25
try to do : r-project¿¡¼ ¸ð¼öÀÇ °ªÀÇ ¹üÀ§¸¦ Á¤ÇÑ ´ÙÀ½ °¢°¢ÀÇ °ª¿¡ ´ëÇÑ divide-and-conquer ½Ãµµ. ¿ÀÈÄ 9:55 2004-06-11, by Kenial
ÆĶó¸ÞÅÍ ¹üÀ§(¿¹»ó)
1 .2a. C(t) : refA, p.25
1 .2b. C(t) : refB, p.86
2 . Ãß»êÇÒ °Í..
4 . C(x,t) : 1.1, 1.2a, 1.2b ·ÎºÎÅÍ Ãß»ê = {1.1}, ({1.2a} + {1.2b}) / 2
5 . C(x,t) : refB : p.88
¿ÀÀü 12:55 2004-06-10, by Kenial
Å×½ºÆ®¿ë ÆĶó¸ÞÅÍ
1-1
C0 .107785647
A 1.305090740
B .074960389
1-2
C0 .191568085
A 1.412937532
B .094437384
2(ÀÌ°Ç Á» ³ªÁß¿¡)
C0 2.255114380
A .259494577
B 8646.0555381
3
C0 .167183448
CX 1.393850202
XX 19.733595112
4
C0 -1.580096488
A 1.471346848
B .088694718
C -1.858608651
D 2.493052423
5
C0 .430609326
CX 3.93449954
XX 8.83168773
T 2.25797694
M -.58709922
¿ÀÀü 10:51 2004-06-08, by Kenial
°ÅÀÇ ¸¶Âù°¡Áö °á·ÐÀÌÁö¸¸, r-project¿¡¼µµ ¸ð¼ö¸¦ ÀûÀýÈ÷ Á¶ÀýÇÏ¸é °°Àº(!) °á°ú¸¦ ¾òÀ» ¼ö ÀÖÀ½.
´ë½Å, ¸ð¼öÀÇ °ª¿¡ ÀÌ»óÀÌ ÀÖÀ» ¶§ ±× ÇØ°áÁ¡À» ã´Â ¹üÀ§¿¡¼ spss°¡ Á» ´õ ¿ì¼ö.
°³°³ÀÇ °ªÀº ¸ð¼ö°¡ °¡Áú ¼ö ÀÖ´Â ÃÖ¼Ò/ÃÖ´ë°ªÀ» 10µîºÐÇÑ °ªÀ¸·Î Çϸç, 10ȸÀÇ ½Ãµµ¿¡ ½ÇÆÐÇϸé
°³°³ÀÇ ¸ð¼ö¸¶´Ù 10´Ü°èÀÇ °ªÀ» ÃëÇØ ¸ð¼ö ÇϳªÇϳªÀÇ °ªÀ» º¯°æÇØ°¡¸ç divide-and-conquer ½Ãµµ
(ÃÖ¾ÇÀÇ °æ¿ì 10^p ¹øÀÇ ½Ãµµ : p=¸ð¼öÀÇ °³¼ö) ¿ÀÀü 1:22 2004-06-08, by Kenial
¸ð¼öparameter Á¶Á¤ ¼º°ø!
spsssyntax_parametered.txt
´ë·« ¾î¹ö¹öÇÏÁö¸¸ À¢°£ÇÑ °á°úÄ¡¿¡¼´Â 4°³ ÀÌ»óÀÇ ¹æÁ¤½ÄÀÌ 80% ÀÌ»óÀÇ r^2 °ªÀ» º¸ÀÓ.
tunnel <- read.table("d:tunnel.txt", TRUE)
³»°øº¯À§ÀÇ °ªÀº ÃøÁ¤Ä¡¿¡ -1À» °öÇØÁÖ¾î¾ß ÇßÀ½.
3¹ø ¸ðµ¨ÀÇ °á°ú :
stableÇÑ °ªÀº ¾Æ´ÏÁö¸¸, ÃʱâÄ¡ÀÇ ¼³Á¤¿¡ µû¶ó Á¦´ë·Î µÈ °ªÀ» ¾ò¾î³¾ ¼ö ÀÖÀ» µí ÇÔ.
¿ÀÀü 1:03 2004-06-07, by Kenial
r-project¿¡¼ÀÇ nonlinear regression ÇÔ¼ö Å×½ºÆ® Áß.
ft1 <- nls( c ~ A * ( 1 - exp( -B * x ) ) - c0, data = tunnel,
algorithm = "plinear",
ft1 <- nls( c ~ (( 1 - exp( -B * x ) ) * A ) - c0, data = tunnel,
start = list( A=0, B=0, c0=0 ), trace = TRUE)
start = list(A=0, B=0, c0=0 ), trace = TRUE)
ft1 <- nls( c ~ (( 1 - exp( -B * t ) ) * A ) - c0, data = tunnel,
start = list(A=0, B=0, c0=0 ), trace = TRUE)
ft1 <- nls( c ~ ( A * log(1 + (B*t)) ) - c0, data = tunnel,
start = list(A=0, B=0, c0=0 ), trace = TRUE)
ft1 <- nls( c ~ ( 1 - ( ( XX / ( XX + x ) ) ** 2 ) ) - c0, data = tunnel,
start = list(XX=0, c0=0 ), trace = TRUE)
ft1 <- nls( c ~ pa * (1 - exp(-pb * x)) + pc * (1 - exp(-pd * t)) - c0, data = tunnel,
start = list(pa=0, pb=0, pc=0, pd=0, c0=0 ), trace = TRUE)
ft1 <- nls( c ~ PCX*(1- ( (PX/(PX+x)) **2 )) * (1+ PM*(1-((PT/(PT+t))**0.3))) - c0,
data = tunnel, start = list(PCX=0, PX=0, PM=0, PT=0, c0=0 ), trace = TRUE)
°á±¹ ¶Ç »ðÁú ÀÛ¾÷À» ÇßÀ½ÀÌ ¹àÇôÁü.
±âÃÊ parameter 0À¸·Î Á¶Á¤ ¾øÀÌ
3¹ø ¸ðµ¨¿¡¼ R^2 = .98826, 5¹ø ¸ðµ¨¿¡¼ R^2 = .91620 ¼öÄ¡ ³ª¿È
(5¹ø ¸ðµ¨ÀÇ °æ¿ì Ç¥ÁØ¿ÀÂ÷°¡ »ó´çÇÏ¿© º° Àǹ̰¡ ¾ø´Â °ª)
Nonlinear Regression Summary Statistics Dependent Variable C
Source DF Sum of Squares Mean Square
Regression 3 10.27649 3.42550
Residual 10 .01351 1.350632E-03
Uncorrected Total 13 10.29000
(Corrected Total) 12 1.15077
R squared = 1 - Residual SS / Corrected SS = .98826
Asymptotic 95 %
Asymptotic Confidence Interval
Parameter Estimate Std. Error Lower Upper
C0 .129819687 .050150412 .018077605 .241561769
CX 1.139746965 .049305817 1.029886758 1.249607172
XX 15.949666320 1.886976217 11.745221299 20.154111341
¿ÀÀü 2:20 2004-06-06, by Kenial
* A * ( 1 - EXP(-B*x) ).
* NonLinear? Regression.
MODEL PROGRAM c0=0 A=0 B=0 .
COMPUTE PRED_ = A * ( 1 - EXP(-B*x) ) - c0.
NLR c
/OUTFILE='C:\SPSSFNLR.TMP'
* A * ( 1 - EXP(-B*t) ).
/PRED PRED_
/CRITERIA SSCONVERGENCE 1E-8 PCON 1E-8 .
* NonLinear? Regression.
MODEL PROGRAM c0=0 A=0 B=0 .
COMPUTE PRED_ = A * ( 1 - EXP(-B*t) ) - c0.
NLR c
Levenberg-Marquardt MethodÀÌ ¾Æ´Ï¶ó¸é... sequential quadratic programmingÀÌ´Ù!
¿ÀÈÄ 4:18 2004-06-03, by Kenial
Levenberg-Marquardt Method¿¡ ´ëÇؼ...
http://groups.google.co.kr/groups?hl=ko&lr=&ie=UTF-8&newwindow=1&threadm=8kjf16%24mnf%241%40b5nntp2.channeli.net&rnum=3&prev=/groups%3Fq%3Dlevenverg-marquardt%2520algorithm%26hl%3Dko%26lr%3D%26ie%3DUTF-8%26newwindow%3D1%26sa%3DN%26tab%3Dwg
http://www-fp.mcs.anl.gov/otc/Guide/OptWeb/index.html
analyze - regression - curve estimation ÈÄ ÇÔ¼öÀÇ Å¸ÀÔÀ» Å×½ºÆ®...!
¿ÀÀü 12:17 2004-06-03, kenial
±×´ÙÀ½ ³ª¿Â b1 °ªÀ» Åä´ë·Î nonlinear regressionÀ» ½Ø¿î´Ù
»ó¼öÇ× Æ÷ÇÔ/ºñÆ÷ÇÔÀ» Âü°í
(ÇÑ±Û spss 10.0¿¡ ÀÇÇÑ ¾Ë±â ½¬¿î ´Ùº¯·®ºÐ¼®, ³ëÇüÁø Àú, Çü¼³ÃâÆÇ»ç) ¿ÀÈÄ 11:45 2004-05-27, kenial
»ùÇà µ¥ÀÌÅ͸¦ ÀÌ¿ë, c = PCX * ( 1- ((PX / (PX + x ))**2)) * ( 1 + PM * (1- (PT / (PT + t))**0.3)) ¹æÁ¤½ÄÀÇ È¸±ÍºÐ¼®À» ½ÃµµÇßÀ¸³ª, PT + t °ªÀÇ 0ÀÏ ¶§°¡ ÀÖ¾î divide by zero ¿¡·¯¸¦ ³»°í »ç¸Á.
PT > a ÀÇ ÀÏÁ¤°ªÀ» ÁöÁ¤ÇÑ ÈÄ Squared R °ªÀÇ º¯È :
a | Squared R |
1E-20 | .04616 |
1E-19 | .04616 |
1E-18 | .56504 |
1E-17 | .56458 |
1E-16 | .56465 |
1E-15 | .56478 |
1E-14 | .56420 |
1E-13 | .56447 |
1E-12 | .56293 |
1E-11 | .56529 |
1E-10 | .56818 |
1E-09 | .59836 |
1E-08 | .62288 |
1E-07 | .77039 |
1E-06 | .63823 |
1E-05 | .74458 |
1E-04 | .89697 |
1E-03 | .89542 |
1E-02 | .89231 |
1E-01 | .89485 |
1 | .89521 |
1E+01 | .85480 |
1E+02 | .62513 |
µµ´ëü ÀÌ °á°ú¸¦ ¾î¶»°Ô ¹Þ¾Æµé¿©¾ß ÇÒÁö ¾Ë ¼ö ¾øÀ½.
¿ÀÈÄ 11:45 2004-05-27, kenial
2Â÷ ¹ÌÆÃ
¿ÀÀü 10:00 2004-05-27, kenial
; r ¾ð¾î ¹× (°³¹ß) ȯ°æ - Åë°è ó¸®¿Í ±×·¡ÇÈ¿¡ ÁßÁ¡À» µÒ. s languageÀÇ gnu ¹öÀüÀ¸·Î ¸¸µé¾îÁ³À¸¸ç(Âü°í:s+´Â s languageÀÇ °³·®ÆÇÀ̸ç, ÀϹÝÀûÀÎ s languageÀÇ ÄÚµå´Â Å« ¼öÁ¤ ¾øÀÌ r-project ȯ°æ¿¡¼µµ »ç¿ë °¡´É) ±âº»ÀûÀÎ s ¾ð¾î¿Í ȯ°æÀ» Á¦°øÇÏ´Â °Í ¿Ü¿¡ contributed packages ¶ó´Â ÇüÅ·Π¿ÜºÎ ÆÐÅ°Áö¸¦ Ãß°¡ÇÔÀ¸·Î¼ ±â´ÉÀ» È®ÀåÇÒ ¼ö ÀÖ´Ù. ±× ¿Ü¿¡µµ ¼ºê ÇÁ·ÎÁ§Æ®µéÀ» °®°í ÀÖÀ¸¸ç, gui ó¸® ȯ°æÀ» Á¦°øÇϰųª graphical modelÀ» ½±°Ô Á¦ÀÛÇÒ ¼ö ÀÖ´Â ¸ñÀûÀÇ ÇÁ·ÎÁ§Æ®, ȤÀº LaTex?¸¦ ÀÌ¿ëÇØ Ãâ·Â°¡´ÉÇÑ ÇüÅÂÀÇ ¹®¼¸¦ ¸¸µé¾î³»´Â ÆÐÅ°Áö µî ¿©·¯ ÇüÅÂÀÇ ¾ÖµåÀÎÀÌ Á¦°øµÊ.
¿ÀÈÄ 6:35 2004-05-26, kenial
1Â÷ ¹ÌÆÃ
; ÀÔ·Â µ¥ÀÌÅÍ Æļ, ¼ö½Ä µ¥ÀÌÅÍ ÀúÀå/Equatation Ãâ·Â/Graph Ãâ·Â Ŭ·¡½º, Regression ÇÁ·Î¼¼½Ì ¸ðµâ
±âŸºÐ·ù WorkBook