Hello!
This is not a real issue but just a suggestion.
I have noticided that EnvCpt is not robust to outliers (see code below). It might be useful to have an additional argument in envcpt() to choose if to perform a robust changepoint analysis or not (See 'robust' argument in the stl() function of the stats R package).
Thank you!
without outlier
library(EnvCpt)
y = c(rnorm(50,100,sd = 5),rnorm(50,50,sd = 5))
out = envcpt(y)
which.min(BIC(out))
plot(out$meancpt,type="p")
with outlier
y[20]=200
out = envcpt(y)
which.min(BIC(out))
plot(out$meancpt,type="p")
Hello! This is not a real issue but just a suggestion. I have noticided that EnvCpt is not robust to outliers (see code below). It might be useful to have an additional argument in envcpt() to choose if to perform a robust changepoint analysis or not (See 'robust' argument in the stl() function of the stats R package). Thank you!
without outlier library(EnvCpt) y = c(rnorm(50,100,sd = 5),rnorm(50,50,sd = 5)) out = envcpt(y) which.min(BIC(out)) plot(out$meancpt,type="p")
with outlier y[20]=200 out = envcpt(y) which.min(BIC(out)) plot(out$meancpt,type="p")