Hi, I'm using the arima function in R to deal with timeseries data, accessing it from an external Java application. The connection to Java though, is not my problem. The problem is that often the arima function does not return a value, but instead throws the error: Error in optim ... initial value is not finite I've searched the net and help listings for a while now, and all I can find is that if the parameters for the ARIMA model are set wrong, then the function cannot create a stable model :-( Does this mean that I have to know beforehand which parameters I need for the model?!! I thought the whole idea was to create a model, see how it behaves, and THEN adapt the parameters. (I know that I can use AutoCorrelation and such to figure out more about the parameters, but anyway.) Thanks Johan Hedlund [[alternative HTML version deleted]]