The RATS Software Forum

This function returns a 2-vector with the parameters for a beta distribution which has a requested mean and standard deviation. The beta is often used for the prior for a model parameter which has support on (0,1) and can be adapted easily to handle support on more general finite intervals.
Because it's a function (not a procedure), you need to do a SOURCE instruction on the file before you can use it in a RATS program.

target VECTOR=%BetaParms(mean,sd)

The mean must in the range (0,1) and the standard deviation in (0,sqrt(mean*(1-mean)))

Example
source betaparms.src (needs to be done once)
compute bparms=%BetaParms(.50,.20)
set test 1 1000 = %ranbeta(bparms(1),bparms(2))
set logg 1 1000 = %logbetadensity(test,bparms(1),bparms(2))

will create a series named TEST with beta draws with (population) mean .5 and standard deviation .2 and the log density (in LOGG) at each drawn value.

Dear RATS users,

is there any simple way to obtain the CDF of a "generalized" beta distribution with support over a to b (rather than 0 to 1) using the basic functions available in RATS?

(I'd like to estimate the parameters of such a distribution based on empirically observed CDF-values from surveys (I have probabilities assigned to different "bins"/forecast intervals) using NLLS.)

Best regards,
Jonas

jonasdovern wrote:Dear RATS users,

is there any simple way to obtain the CDF of a "generalized" beta distribution with support over a to b (rather than 0 to 1) using the basic functions available in RATS?

(I'd like to estimate the parameters of such a distribution based on empirically observed CDF-values from surveys (I have probabilities assigned to different "bins"/forecast intervals) using NLLS.)

Best regards,
Jonas

With support on [a,b] and shape parameters alpha,beta:

compute cdfgbeta=%betainc((x-a)/(b-a),alpha,beta)