In reading and experimenting with numpy.random, I can't seem to find or create what I need; a 10 parameter Python pseudo-random value generator including count, min, max, mean, sd, 25th%ile, 50th%ile (median), 75th%ile, skew, and kurtosis.
From https://docs.python.org/3/library/random.html I see these distributions uniform, normal (Gaussian), lognormal, negative exponential, gamma, and beta distributions, though I need to generate values directly to a distribution defined only by my 10 parameters, with no reference to a distribution family.
Is there documentation, or an author(s), of a numpy.random.xxxxxx(n, min, max, mean, sd, 25%, 50%, 75%, skew, kurtosis), or what please is the closest existing source code that I might modify to achieve this goal?
This would be the reverse of describe() including skew and kurtosis in a way. I could do a loop or optimize until a criteria is met with random generated numbers, though that could take an infinite amount of time to meet my 10 parameters.
I have found optim in R which generates a data set, but have so far been able to increase the parameters in the R optim source code or duplicate it with Python scipy.optimize or similar, though these still depend on methods instead of directly psudo-randomly creating a data set according to my 10 parameters as I need to;
m0 <- 20
sd0 <- 5
min <- 1
max <- 45
n <- 15
set.seed(1)
mm <- min:max
x0 <- sample(mm, size=n, replace=TRUE)
objfun <- function(x) {(mean(x)-m0)^2+(sd(x)-sd0)^2}
candfun <- function(x) {x[sample(n, size=1)] <- sample(mm, size=1)
return(x)}
objfun(x0) ##INITIAL RESULT:83.93495
o1 <- optim(par=x0, fn=objfun, gr=candfun, method="SANN", control=list(maxit=1e6))
mean(o1$par) ##INITIAL RESULT:20
sd(o1$par) ##INITIAL RESULT:5
plot(table(o1$par))
Aucun commentaire:
Enregistrer un commentaire