1

I am trying to find my way around in the different R time series packages.

My data series consists of hourly prices for around 7 years. The rows are the days, and every row has 24 hourly values.

I know that there are daily effects, weekly effects and longer seasonal effects. I want to run forecasting models like ARIMA. However most models (and ts objects) only allow a single frequency for seasonal effects. This is for example the different hourly averages per month.

enter image description here

Which models can I use, and what form should I store my data in to use them?

Kasper Van Lombeek
  • 623
  • 1
  • 7
  • 17
  • What is the question? – zx8754 Feb 14 '14 at 14:42
  • `melt` (from package reshape2) it? Hard to say without more details. – Roland Feb 14 '14 at 14:48
  • 1
    I have voted to close. Please refer to point 1 & 5 [**here**](http://stackoverflow.com/help/on-topic). – Henrik Feb 14 '14 at 14:58
  • How is this not a good question? I am asking what is the best practice regarding handling such a matrix, there are soo many options that I think it is wise to ask on this forum. – Kasper Van Lombeek Feb 14 '14 at 15:00
  • 1
    Why don't you try to describe one of the problems you're trying to solve with this type of data, and provide a small sample of the data (see [reproducible examples](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example)) so people can give you concrete suggestions? – BrodieG Feb 14 '14 at 15:12
  • In addition to the link I posted above and the one provided by @BrodieG, you may have a look at these nice links for ideas on how to improve your question: [**here**](http://meta.stackoverflow.com/help/how-to-ask), [**here**](http://meta.stackexchange.com/questions/156810/stack-overflow-question-checklist). Cheers. – Henrik Feb 14 '14 at 15:15
  • I'm also voting to close due to subjectivity. The best form for the data depends on what you want to do with it. As others have said: provide a concrete example of a problem that you need solving; otherwise there is no way to sensibly answer this question. – Richie Cotton Feb 14 '14 at 15:41
  • Question edited with clear goals of my data analysis – Kasper Van Lombeek Feb 14 '14 at 15:48
  • @KasperVanLombeek: I've rewritten what I think you want, in a way that shouldn't be closed or ignored. Obviously feel free to change it if this isn't what you want. – Richie Cotton Feb 14 '14 at 16:37
  • @RichieCotton That is indeed a much clearer question, next time I'll think twice, thank you. BTW, I started in R already with one long vector, the advantage I think is that it is very easy to make dummie colomns. – Kasper Van Lombeek Feb 14 '14 at 16:50

0 Answers0