Error in PROC AUTOREG.

[damanaulakh88]

Can you please help me on this below error:-

WARNING: Exact fit. No residual in model to analyze.

NOTE: The above message was for the following BY group:

      Terminal=Hamar postterminal prodgrservniva=10 prodgrformat=Små

WARNING: Exact fit. No residual in model to analyze.

NOTE: The above message was for the following BY group:

      Terminal=Hamar postterminal prodgrservniva=10 prodgrformat=Store

WARNING: Exact fit. No residual in model to analyze.

NOTE: The above message was for the following BY group:

      Terminal=Hamar postterminal prodgrservniva=10 prodgrformat=Klump

ERROR: Floating Point Overflow.

ERROR: Termination due to Floating Point Exception

NOTE: The SAS System stopped processing this step because of errors.

Below is the code i am using:-

PROC AUTOREG DATA=prognose noprint;

        by terminal prodgrservniva prodgrformat;

        MODEL ant_sendinger=

        man tir ons tor fre lor son

        rompaske1 rompaske2 rompaske3 romjul

        fellesferie skoleferie

        hel_for_arb jul_ny_aften helg_f_jul

        arb_etter_hel trend1 trend2

        /NLAG=2 NOINT method=ml

                                ;

        OUTPUT OUT=res_prognose

        p=antallhat pm=antalltrend r=residind lcl=lcl ucl=ucl alphacli=0.20;

      RUN;

Thanks,

Daman

[SteveDenham]

Without greater information, I would assume the ERROR message is telling me that the model overfits the data for those by groups.  I would guess that there are 20 or fewer observations for those groups, thus resulting in an exact fit, and no residual error.  Without additional info, this is as close a guess as I can hazard.

Steve Denham

[JonathanW]

agreed. When I saw the list of independ variables, the first thing in mind is possible over-fitting. It might be more than over-fitting in his case.