Example is always good to include in your questionâ¦
Anyways, you do not just take logarithm. First of all you want time series to be stationary. This can be achieved in 2 ways: 1. Fit the trend (fit line, polynomial or anything you want) and seasonality (if exists) 3. Filtering 2. First differencing Then do your ARMA analysis of residuals. If we are talking about stock prices, then you can not really fit the trend for any forecasting purposes as stock price is a random process meaning that trend will not tell much. For financial time series standard procedure is log-returns. It is not simple logarithm, it is differences of logarithms. Kind regards,-- Dominykas Grigonis On Thursday, 15 May 2014 at 19:32, Christofer Bogaso wrote: > Hi again, > > My question is not directly related to R, but I still hope to get some > quality feedback on my question. > > In a typical time series analysis, we generally take logarithm of the > underlying variable and then fit models like ARIMA etc on that transformed > variable. This has a related benefit to getting linear trend for model > variable instead of quadratic > > This essentially assumes that, the variable under study is strictly > positive. However what if my variable is mix of positive, zero, and > negative? would it then make sense to fit model on the raw variable without > transforming it anyway? > > Is anybody aware of any standard policy to handle such situation? > > Thanks for your input. > > [[alternative HTML version deleted]] > > _______________________________________________ > R-SIG-Finance@r-project.org (mailto:R-SIG-Finance@r-project.org) mailing list > https://stat.ethz.ch/mailman/listinfo/r-sig-finance > -- Subscriber-posting only. If you want to post, subscribe first. > -- Also note that this is not the r-help list where general R questions > should go. > > [[alternative HTML version deleted]]
_______________________________________________ R-SIG-Finance@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.
