I’m looking for some procedures to simulate time series based on another time series as an input.

The objective is that it may provide me with more training datasets that are similar to the given time series.

A few methods that come to my mind is by using GAN(generative adversarial network) and Monte Carlo.

Ideally there’re some R packages that can help with generating the time series simulation.

Open to ideas, thanks!

  • $\begingroup$ Perhaps the discussion of time series bootstrapping at stats.stackexchange.com/questions/25706 will give you some useful ideas. $\endgroup$
    – whuber
    Apr 17 '20 at 20:49
  • $\begingroup$ @whuber thanks, reading it now :) $\endgroup$ Apr 18 '20 at 6:41

"based on another time series as an input" means to me to identify and estimate a useful model using a user-specified time series. Find a piece of software that allows you to specify a particular with specified coefficient values and use that model as the basis for a monte-carlo simulation. I only know one piece of software that allows this gambit , it is called AUTOBOX which I have helped to develop but there might be others . Perhaps other readers of this list can specify alternative softwares as I would be interested in evaluating/researching them.

  • $\begingroup$ Thanks for the Autobox reference, will check it out $\endgroup$ Apr 18 '20 at 6:42
  • $\begingroup$ when and if you investigate this feature , please contact me offline and I will try and help you further.. I believe it is a unique feature of AUTOBOX. $\endgroup$
    – IrishStat
    Apr 18 '20 at 9:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.