R help - Mar 2014 - how to do ARIMA prediction with high performance

Hi,


I need to process a data stream using ARIMA model in real-time, that is : I need
to build a ARIMA model and do prediction in every 1 minute using a slide window
with 200 samples .


To improve the performance of prediction, what I plan to do is: 1) to build
ARIMA model using some history data;
2) to predict the data within slide window using the ARIMA got in step 1) .


Is it possible to do this? and how?


Or is there other way to do the  prediction with high performance?


Thank you very much!


Hellen
	[[alternative HTML version deleted]]