can I develop a forecast model with time interval of every 5 minutes (sas ets 9.4)?

[Hansdewit]

I receive  time serie data with an interval of 5 minutes.  we have in our company in sas 9.4 environment, Sas ETS.  I am using at this moment proc ucm.

can i use an interval of every 5 minutes? or should I change the interval to minutes or hours?

trying to detect some outliers.

[Ksharp]

Did you try PROC ARIMA ?
Check its example :
Example 8.7: Iterative Outlier Detection

[Hansdewit]

i know I can use proc arima as well. but my question if I can use 5 minutes interval data.  it looks like I can only choose 1 minute data or 1 hour interval. 

[Ksharp]

Sorry. I am not expert about your question. Maybe some other sas user could give you an hand.

[mitrov · Accepted solution]

Yes, you can do that. In the ID statement of PROC UCM, you need to specify a datetime variable that contains the timestamp and a corresponding interval of 5 minutes. You can use a multiplier or a shift to customize the interval to your need. In your case it should be INTERVAL=MINUTE5.

That said, you need to think about whether you really need to forecasts at 5 minutes intervals. Are decisions based on forecasts taken at 5 minutes intervals? If not, maybe you need to think about the best interval to use. 

[Hansdewit]

our purpose is to identify anomalies during the day on our mobile network.  we will run the model every day, forecast the next day.  if the real cases are outside the boundaries it is anomaly, and we will inform some decision makers.