I want to make a thing that produces a stochastic process from time series data. The time series data is recorded every hour over the year, which means 24-hour of patterns exist for 365 days.

What I want to do is something like below:

  1. Fit a probability distribution using data for each hour so that I can sample the most probable value for this hour.

  2. Repeat it for 24 hours to generate a pattern of a day.

BUT! I want the sampling to be done considering previous values rather than being done in an independent manner.

For example, I want to sample from p_{h1}(x_t|x_{t-1}, x_{t-2}, ... , x_{t-k}) or just p_{h1}(x_t|x_{t-1}) rather than p_{h1}(x_t) when h1 refers to a specific hour.

What I came up with was the Markov chain, but I couldn't find any reference or materials on how to model it from real data.

Could anyone give me a comment for this issue?


Your Answer

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

Browse other questions tagged or ask your own question.