Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 1 down vote favorite

I'm forecasting gold prices using an ARIMA model. An ARIMA model requires a stationary, non-seasonal, linear series. However, after reading a few books, it seems that gold price data is nonstationary, seasonal, and non linear.

Can anyone please give me some suggestions and solutions for the question?

share|improve this question
ARIMA models do not need stationary and non-seasonal data. They are linear. Gold data is certainly not seasonal (or people would exploit the seasonality to make money). It is non-stationary, and it is debatable whether it is linear. – Rob Hyndman Dec 30 '11 at 11:55

1 Answer

up vote 4 down vote

ARIMA can handle seasonality, depending on the implementation in your software of choice. The "I" part is to take care of certain kinds of non-stationarity, though other kinds would require a de-trending pre-processing step of some sort.

EDIT: In answer to your comment, I'd recommend you look into the Box-Jenkins methodology. It's a series of steps to analyze your data to determine what preprocessing you need and what are the proper ARIMA parameters. Unfortunately, it's a bit of an art that you have to learn. If you're using a statistical package, some of them have ARIMA methods that do some of this analysis for you and make suggestions.

Taking a small step back, almost all data has to be preprocessed in some manner before use. (If nothing else, looking for errors and issues.)

Taking a large step back, I'd warn that you're not going to predict gold or stock prices accurately. A lot of people have tried very hard, inventing and using much more advanced techniques than ARIMA (which is very basic, and fairly old). I hope this is just an academic exercise to learn ARIMA, without any expectation that you're going to actually invest any money.

share|improve this answer
does it mean i no need to make any changes in my data before i key in the software? – Morae Dec 29 '11 at 14:44
1  
+1 for the large step back – RockScience Dec 30 '11 at 4:30
Thanks for all the replies, really helpful indeed, actually im doing it as my final year project, comparing ARIMA, Fuzzy Time Series and ANFIS ^^ – Morae Dec 31 '11 at 14:06

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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