topic Re: Scoring a data using the coefficients of proc transreg in Statistical Procedures

Scoring a data using the coefficients of proc transreg

roushankumar — Mon, 17 Sep 2018 18:48:10 GMT

Hi,

I created a model of the following form

proc transreg data=mydata outest=work.est;

model BoxCox(Sales) = identity ( x1 x2 x3 x4 );

run;

When I check the table work.est I do not see coefficients against x1 x2 x3 and x4. The column 'Variable' in work.est does not have any values. I do see columns 'Statistic' and 'Value' which have Error, Root RMSE etc. How do I score an external data using this model's coefficients? Since I have used a boxcox transformation, what transformation do I need to get the predictions in the original form? Thanks for the help!

Re: Scoring a data using the coefficients of proc transreg

Rick_SAS — Fri, 21 Sep 2018 14:18:00 GMT

All SAS regression procedure support the "missing value trick" for scoring external data. I don't use PROC TRANSREG enough to know whether it supports other scoring options.

Re: Scoring a data using the coefficients of proc transreg

roushankumar — Mon, 24 Sep 2018 14:13:30 GMT

Hi Rick,

Could you please help me with the question in this link?

https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/Exception-with-Proc-Forecast-when-specifying-SEASONS-MONTH/m-p/498094

Thanks,

Roushan