calculate Standard Error of parameter of Beta Function
-
21-12-2019 - |
Question
I'm learning R using a book called "Discovering Statistics using R". It's great but seems to skip over some areas.
So, I have a function in R that calculates the parameters a, b
of the following pdf beta function:
My function returns the following parameters found from a sample of 500:
[1] 1.028316 2.095143 #a b
I'm trying to calculate the standard error of the parameters.
I'm wondering how this can be implemented in R?
As far as I can find online, standard errors are calculated from the sample, not the parameters. So, all i've implemented is, here:
stderr <- function(x) sqrt(var(x)/length(x))
Thanks in advance.
Solution
I'm not sure which function you're using to fit a beta distribution to your data, but the fitdistr
function from the MASS
package provides standard error values for the shape1
and shape2
parameters of the beta distribution:
# Obtain data to fit
set.seed(144)
data <- rbeta(500, 1, 2)
# Fit and output result
library(MASS)
fit <- fitdistr(data, "beta", start=list(shape1=0.5, shape2=0.5))
fit
# shape1 shape2
# 1.0596902 2.0406073
# (0.0602071) (0.1284133)
Here, the standard errors are 0.060 for shape1
and 0.128 for shape2
. You can get at the values with fit$sd
.