set.seed(123)
Replicas = replicate(5000, rnorm(120,3,0.8))
class(Replicas)
## [1] "matrix" "array"
dim(Replicas)
## [1] 120 5000
Replicas_trunc_80 = matrix(nrow = 80, ncol = 5000)
Replicas_trunc_60 = matrix(nrow = 60, ncol = 5000)
Replicas_trunc_30 = matrix(nrow = 30, ncol = 5000)
for(i in 1:dim(Replicas)[2]){
r1 = sort(Replicas[,i])
Replicas_trunc_80[,i] = r1[21:100]
Replicas_trunc_60[,i] = r1[31:90]
Replicas_trunc_30[,i] = r1[46:75]
}
SR = cov(Replicas)
ST_80 = cov(Replicas_trunc_80)
ST_60 = cov(Replicas_trunc_60)
ST_30 = cov(Replicas_trunc_30)
Det_SR = det(SR)
Det_ST_80= det(ST_80)
Det_ST_60 = det(ST_60)
Det_ST_30 = det(ST_30)
#Como el determinante de la matriz de varianzas covarianzas, es decir la varianza generalizada, de los datos truncados es menor, se observa que tienen un menor grado de dispersion en el espacio, al aumentar el numero de datos que son eliminados, se observa que los datos dejan de ser independientes.
min_t_80 = c(); max_t_80 = c()
min_t_60 = c(); max_t_60 = c()
min_t_30 = c(); max_t_30 = c()
min_r = c(); max_r = c()
for(i in 1:dim(Replicas)[2]){
min_r[i]= min(Replicas[,i])
max_r[i]=max(Replicas[,i])
min_t_80[i]=min(Replicas_trunc_80[,i])
max_t_80[i]=max(Replicas_trunc_80[,i])
min_t_60[i]=min(Replicas_trunc_60[,i])
max_t_60[i]=max(Replicas_trunc_60[,i])
min_t_30[i]=min(Replicas_trunc_30[,i])
max_t_30[i]=max(Replicas_trunc_30[,i])
}
Cor_R = cor(min_r, max_r)
Cor_80 = cor(min_t_80,max_t_80)
Cor_60 = cor(min_t_60,max_t_60)
Cor_30 = cor(min_t_30,max_t_30)
par(mfrow=c(2,2))
plot(min_r,max_r, pch=19, cex=0.5)
plot(min_t_80,max_t_80, pch=19, cex=0.5)
plot(min_t_60,max_t_60, pch=19, cex=0.5)
plot(min_t_30,max_t_30, pch=19, cex=0.5)
Determinantes = c(Det_SR, Det_ST_80, Det_ST_60, Det_ST_30)
Truncado = c("120D", "80D", "60D", "30D")
Corr = c(Cor_R, Cor_80, Cor_60, Cor_30)
data.frame(Truncado, Determinantes, Corr)
## Truncado Determinantes Corr
## 1 120D 0 -0.004447881
## 2 80D 0 0.190773361
## 3 60D 0 0.329526214
## 4 30D 0 0.616253984