Contents

1
2
3 るジ 覦覯
4 谿瑚襭


1 #

  • 覲蠏碁9(X={x1,x2, x3}, Y={y1,y2,y3}) 螳 蟯螻
  • 企 譯殊れ 覦襦 "譯殊焔覿 + 蠏覿"

2 #

tmp <- textConnection( 
  "tv    蠍磯レ	襷るル	碁
  1	46	34	28	39
  2	60	31	50	46
  3	81	59	63	72
  4	94	84	92	92
  5	76	67	86	52
  6	31	53	41	39
  7	34	38	25	25
  8	78	75	64	76
  9	54	43	38	55
  10	86	53	60	70
  11	53	43	34	42
  12	78	31	52	67
  13	96	66	77	88
  14	71	90	86	65
  15	67	58	60	70
  16	32	68	74	45
  17	44	55	60	42
  18	59	46	42	67
  19	76	30	37	64
  20	84	51	54	79")
x <- read.table(tmp, header=TRUE) 
close.connection(tmp)
#head(x)

library("sqldf")
d1 <- sqldf("select , 蠍磯レ from x")
d2 <- sqldf("select 襷るル, 碁 from x")
rs1 <- cancor(d1, d2)
rs1$cor

> X <- with(x, -0.007865095 * (-65.00) + -0.006951716 * (蠍磯レ-53.75))
> Y <- with(x, -0.007865095 * (襷るル-56.15) + -0.006951716 * (碁-59.75))
> cor.test(X,Y)

	Pearson's product-moment correlation

data:  X and Y
t = 13.0087, df = 18, p-value = 1.362e-10
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 0.8772728 0.9806609
sample estimates:
      cor 
0.9507151 

> plot(X,Y)

#install.packages("CCA")
library("CCA")
rs2 <- cc(d1, d2)
plot(rs2$scores$xscores[,1], rs2$scores$yscores[,1])

3 るジ 覦覯 #

#install.packages("yacca")
library("yacca")
rs3 <- cca(d1, d2)
rs3

> rs3

Canonical Correlation Analysis

Canonical Correlations:
     CV 1      CV 2 
0.9558493 0.6976745 

X Coefficients:
              CV 1        CV 2
 -0.03428316  0.03920114
蠍磯レ -0.03030183 -0.05234068

Y Coefficients:
              CV 1        CV 2
襷るル -0.02577813 -0.05651435
碁 -0.03411328  0.05890991

Structural Correlations (Loadings) - X Vars:
             CV 1       CV 2
 -0.8654311  0.5010280
蠍磯レ -0.7527465 -0.6583105

Structural Correlations (Loadings) - Y Vars:
             CV 1       CV 2
襷るル -0.8653783 -0.5011192
碁 -0.9098212  0.4150005

Aggregate Redundancy Coefficients (Total Variance Explained):
	X | Y: 0.7675629 
	Y | X: 0.823285

> plot(rs3)

伎 企慨覃...
1.png
  • 1譴蟯覲(CV1) - 覲蠏碁9X 覲蠏碁9Y 豕 蟯 螻 0.96
  • 2譴蟯覲(CV2) - 覲蠏碁9X 覲蠏碁9Y 豕 蟯 螻 れ朱 蟯螻 0.7
Canonical Correlations:
     CV 1      CV 2 
0.9558493 0.6976745 

2.png
Aggregate Redundancy Coefficients (Total Variance Explained):
	X | Y: 0.7675629 
	Y | X: 0.823285


3.png

Structural Correlations (Loadings) - X Vars:
             CV 1       CV 2
 -0.8654311  0.5010280
蠍磯レ -0.7527465 -0.6583105

Structural Correlations (Loadings) - Y Vars:
             CV 1       CV 2
襷るル -0.8653783 -0.5011192
碁 -0.9098212  0.4150005
  • 譴(canonical loadings) - 蟆讀 覦覯
  • X蠏碁9 覲 , 蠍磯レ炎骸 CV1, CV2 蟯螻
  • Y蠏碁9 覲 襷るル, 碁 CV1, CV2 蟯螻
  • 谿瑚: 譴蟲谿(canonical cross-loadings, 覲 企 覲蠏碁9 觜蟲蠏碁9 譴蟯覲 蟯蟯螻)朱 蟆 .

4.png
X Coefficients:
              CV 1        CV 2
 -0.03428316  0.03920114
蠍磯レ -0.03030183 -0.05234068

Y Coefficients:
              CV 1        CV 2
襷るル -0.02577813 -0.05651435
碁 -0.03411328  0.05890991

譴蟯螻

  • CV1
    • X = -0.03428316 * + -0.03030183 + 蠍磯レ
    • Y = -0.02577813 * 襷るル + -0.03411328 * 碁
  • CV2
    • X = 0.03920114 * + -0.05234068 + 蠍磯レ
    • Y = -0.05651435 * 襷るル + 0.05890991 * 碁

> X <- with(x, -0.03428316 *  + -0.03030183 * 蠍磯レ)
> Y <- with(x, -0.02577813 * 襷るル + -0.03411328 * 碁)
> cor.test(X,Y)

	Pearson's product-moment correlation

data:  X and Y
t = 13.8003, df = 18, p-value = 5.156e-11
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 0.8896248 0.9827032
sample estimates:
      cor 
0.9558493 

企 譴蟯螻蟾讌 碁? (Bartlett's test)
> F.test.cca(rs3)

	F Test for Canonical Correlations (Rao's F Approximation)

         Corr        F   Num df Den df    Pr(>F)    
CV 1  0.95585 30.00045  4.00000     32 2.029e-10 ***
CV 2  0.69767 16.12224  1.00000     17 0.0008971 ***
---
Signif. codes:  0 *** 0.001 ** 0.01 * 0.05 . 0.1   1 
  • 1譴蟯覲, 2譴蟯覲 覈 碁.



4 谿瑚襭 #