Foram testados 6 grupos de treinamento com 100, 200, 400, 600, 800 e 1000 instâncias respectivamente.
Cada clusterização foi repetida 10 vezes e a entropia para cada experimento foi calculada usando a função do R entropy.ChaoShen().
fclust = read.csv("summaryClusterAssign.csv", header = FALSE)
library(entropy)
ent = apply(fclust, 1, function(x) entropy.ChaoShen(x))
fclust$ent <- ent
boxplot(at = c(1, 2, 4, 6, 8, 10), boxwex = 0.8, fclust$ent ~ fclust$V1, ylim = c(0.5,
2), xlab = "Number of instances by training group", ylab = "Shennon Entropy")
medians = c(median(fclust$ent[fclust$V1 == 100]), median(fclust$ent[fclust$V1 ==
200]), median(fclust$ent[fclust$V1 == 400]), median(fclust$ent[fclust$V1 ==
600]), median(fclust$ent[fclust$V1 == 800]), median(fclust$ent[fclust$V1 ==
1000]))
lmed = lm(medians ~ c(100, 200, 400, 600, 800, 1000))
abline(lmed, col = "red")
summary(lmed)
##
## Call:
## lm(formula = medians ~ c(100, 200, 400, 600, 800, 1000))
##
## Residuals:
## 1 2 3 4 5 6
## -0.013454 0.008283 -0.000125 0.007503 0.012779 -0.014987
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.33e+00 1.02e-02 129.55 2.1e-08 ***
## c(100, 200, 400, 600, 800, 1000) -1.98e-05 1.69e-05 -1.17 0.31
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.0132 on 4 degrees of freedom
## Multiple R-squared: 0.255, Adjusted R-squared: 0.0692
## F-statistic: 1.37 on 1 and 4 DF, p-value: 0.306
abline(lmed,col='red')
100,7,29,31,33
100,6,15,17,62
100,12,21,32,35
100,10,25,31,34
100,5,21,34,40
100,8,16,24,52
100,9,25,29,37
100,14,17,22,47
100,3,27,29,41
100,6,7,23,64
200,16,43,59,82
200,25,36,66,73
200,15,46,68,71
200,32,36,54,78
200,14,21,48,117
200,40,42,56,62
200,41,44,54,61
200,29,35,48,88
200,6,27,33,134
200,1,19,70,110
400,62,73,119,146
400,28,66,76,230
400,53,63,123,161
400,16,92,119,173
400,39,44,79,238
400,45,73,132,150
400,23,101,136,140
400,27,104,127,142
400,32,83,128,157
400,31,57,72,240
600,104,108,120,268
600,35,55,158,352
600,62,121,134,283
600,54,106,178,262
600,68,104,175,253
600,66,100,107,327
600,102,104,178,216
600,61,127,197,215
600,42,104,186,268
600,48,178,186,188
800,58,78,245,419
800,115,146,162,377
800,68,108,247,377
800,106,198,247,249
800,48,236,255,261
800,111,222,225,242
800,47,160,260,333
800,110,155,167,368
800,87,167,249,297
800,32,67,232,469
1000,96,215,296,393
1000,65,103,302,530
1000,28,68,325,579
1000,72,141,146,641
1000,131,165,313,391
1000,54,192,309,445
1000,83,212,337,368
1000,105,114,309,472
1000,43,237,274,446
1000,82,266,314,335