Sur mon dendrogramme, il reste sous le titre la mention " hclust (*, "complete") et je ne sais pas comment la supprimer ?
(j'ai utilisé la fonction rest.hclust).
Merci
Thierry
- > read.table("cla88.txt")
denv denm denc ouve
1 244 3 7 9
2 252 4 5 9
3 216 4 6 9
4 285 3 6 9
5 264 3 4 9
6 281 5 2 77
7 274 2 4 77
8 268 1 6 77
9 528 12 22 9
10 642 25 23 9
11 295 2 1 9
12 198 2 5 9
13 185 4 2 18
14 175 2 2 18
15 144 1 2 18
16 199 1 2 18
17 201 5 3 9
18 210 6 3 9
19 222 7 4 9
20 224 6 6 9
> t8<-read.table("cla88.txt")
> attach(t8)
The following objects are masked from t8 (pos = 3):
denc, denm, denv, ouve
> t8
denv denm denc ouve
1 244 3 7 9
2 252 4 5 9
3 216 4 6 9
4 285 3 6 9
5 264 3 4 9
6 281 5 2 77
7 274 2 4 77
8 268 1 6 77
9 528 12 22 9
10 642 25 23 9
11 295 2 1 9
12 198 2 5 9
13 185 4 2 18
14 175 2 2 18
15 144 1 2 18
16 199 1 2 18
17 201 5 3 9
18 210 6 3 9
19 222 7 4 9
20 224 6 6 9
> summary(t8)
denv denm denc ouve
Min. :144.0 Min. : 1.00 Min. : 1.00 Min. : 9
1st Qu.:200.5 1st Qu.: 2.00 1st Qu.: 2.00 1st Qu.: 9
Median :234.0 Median : 3.50 Median : 4.00 Median : 9
Mean :265.4 Mean : 4.90 Mean : 5.75 Mean :21
3rd Qu.:275.8 3rd Qu.: 5.25 3rd Qu.: 6.00 3rd Qu.:18
Max. :642.0 Max. :25.00 Max. :23.00 Max. :77
> pa<-rpart(formula = ouve ~ ., data = t8, control = rpart.control(minsplit = 6))
> summary(pa)
Call:
rpart(formula = ouve ~ ., data = t8, control = rpart.control(minsplit = 6))
n= 20
CP nsplit rel error xerror xstd
1 0.49008891 0 1.00000000 1.184964 0.5079203
2 0.01982217 2 0.01982217 2.031964 0.6564241
3 0.01000000 3 0.00000000 1.842503 0.6353454
Variable importance
denv denc denm
60 20 20
Node number 1: 20 observations, complexity param=0.4900889
mean=21, MSE=565.8
left son=2 (13 obs) right son=3 (7 obs)
Primary splits:
denv < 266 to the left, improve=0.27967790, (0 missing)
denm < 2.5 to the right, improve=0.12121320, (0 missing)
denc < 6.5 to the right, improve=0.04491298, (0 missing)
Surrogate splits:
denm < 9.5 to the left, agree=0.75, adj=0.286, (0 split)
denc < 14.5 to the left, agree=0.75, adj=0.286, (0 split)
Node number 2: 13 observations, complexity param=0.01982217
mean=11.76923, MSE=17.25444
left son=4 (9 obs) right son=5 (4 obs)
Primary splits:
denc < 2.5 to the right, improve=1.0000000, (0 missing)
denv < 200 to the right, improve=0.7111111, (0 missing)
denm < 1.5 to the right, improve=0.4090909, (0 missing)
Surrogate splits:
denv < 191.5 to the right, agree=0.923, adj=0.75, (0 split)
denm < 1.5 to the right, agree=0.846, adj=0.50, (0 split)
Node number 3: 7 observations, complexity param=0.4900889
mean=38.14286, MSE=1132.408
left son=6 (4 obs) right son=7 (3 obs)
Primary splits:
denv < 283 to the right, improve=1.0, (0 missing)
denm < 8.5 to the right, improve=0.3, (0 missing)
denc < 14 to the right, improve=0.3, (0 missing)
Surrogate splits:
denm < 2.5 to the right, agree=0.714, adj=0.333, (0 split)
denc < 5 to the right, agree=0.714, adj=0.333, (0 split)
Node number 4: 9 observations
mean=9, MSE=0
Node number 5: 4 observations
mean=18, MSE=0
Node number 6: 4 observations
mean=9, MSE=0
Node number 7: 3 observations
mean=77, MSE=0
> plot(pa, uniform=TRUE)
> text(pa)
> library(stats)
> hc<-hclust(dist(t8)
+ )
> hc<-hclust(dist(t8))
> plot(hc)
> hc$order
[1] 9 10 6 7 8 4 11 5 1 2 16 12 17 19 20 3 18 15 13 14
> rect.hclust(hc,k5,border="red",which=c(1,2,3,4))
Erreur dans rect.hclust(hc, k5, border = "red", which = c(1, 2, 3, 4)) :
objet 'k5' introuvable
> rect.hclust(hc,k52,border="red",which=c(1,2))
Erreur dans rect.hclust(hc, k52, border = "red", which = c(1, 2)) :
objet 'k52' introuvable
> rect.hclust(hc,k2,border="red",which=c(1,2))
Erreur dans rect.hclust(hc, k2, border = "red", which = c(1, 2)) :
objet 'k2' introuvable
> rect.hclust(hc,k=5,border="red",which=c(1,2,3,4))
> rect.hclust(hc,k=5,border="red",which=c(1,2,3,4,5))
> save.image("G:\\Test 8\\TestsPartyRpart_ok.RData")
> plot(hc, main="Dendrogramme Test sur JDD partiel Marion")
> plot(hc, main="Dendrogramme sur JDD de Marion",xlab=" Sites", ylab="Hauteur")
> rect.hclust(hc,k=5,border="red",which=c(1,2,3,4,5))
> save.image("G:\\Test 8\\TestsPartyRpart_Ok2.RData")
>