Hi every body
I want to simulate data and analyze them for 1000 times, then I need all 1000 dataset that are simulated in order to analyze them with another approach and compering the results of two approaches, BUT I don't know how to save dataset
I see two flavors of the problem here:
1. How to save the results of simulating software ('stata'), not only see it on the screen, or
2. What is the input data format (organization) recognizable by the "other software?"
In both cases familiarization with appropriate manuals seems unavoidable.
But, maybe, you are asking whether 1000 results should be stored as a single big disk file or as 1000 individual files? In the second case you probably will be better of to save all files in a separate disk path, with individual files having names ending with something like '0000.dat', '0001.dat', and so on. In the first case you have to be able to append new data series to an existing file, adding some extra separator (empty line?) between successive series.
You can save the estimates from each simulation by using the save() option. For example, let's say you created a program called mysim. Then you run the simulate command as follows:
simulate mean=r(mean) var=r(Var), reps(10000) saving(sim, replace): mysim, obs(100)
A file will be generated and saved called "sim" that will have the 10,000 mean and variances produced from the 10,000 simulations.
Type: help simulate
I hope this helps
Ariel
dear Ariel Linden
Thank you very much
but I need to save each dataset that for example mean and var are calculated from them
You want to save 10,000 datasets??? Why? The simulations should give you the info you seek so that you don't need to save all the datasets!
In any case, you could write your simulation code to save the file after creating it within a loop. At the end of each loop write:
save sim`i' (you need to specify what folder to save these files under)
dear Ariel Linden
Exactly I want to save all 10000 dataset in order to analyze them with other approach and compering the results
Thanks alot
Dear Ariel Linden
I have done your suggestion BUT it seems simulate() function does not use that dataset!
for example the mean of x in my first dateset (sim1.dta) is 2 but in first line in simulation program which must be corresponded to first dataset is 9!!
thsanks
Dear Marek Wojciech Gutowski
Thanks for your suggestion but I could n't use it
Amir, you need to save the data after generating it in the data generating process, not at the stage you implement simulate. When you save the results from simulate, then those are the summary results stored in r().
Well, the old saying "when all else fails it's a time to see the instructions" seems still valid.
Thanks both of you
suppose I have written this simple program in order to simulate mean of x variable:
set seed 123
program drop _all
program define m1
local r=20
forvalues i= 1/`r'{
drop _all
set obs 100
gen x=rnormal(1,7)
summarize x
scalar meanx= r(mean)
save "D:\stata\mean`i'.dta", replace
}
end
m1
simulate mean_x=meanx,reps(200) :m1
By loop and term "save "D:\stata\mean`i'.dta", replace" ,,datasets are saved.
But they are not that dataset which simulate() function uses, I need exactly that datasets which have been used by simulate() function
This code looks fine. The simulation is running on the data above. Run the simulation with 1 rep. It should give you the same results.
Dear Ariel Linden
Simulation with 1 rep is great, and I can get the dataset, BUT surprisingly with simulation more than 1, result is NOT the same!
Dear Leila Janani
almost all of my simulation codes have been written by STATA.
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators