https://communities.sas.com/t5/Statistical-Procedures/Proc-GAM/m-p/35376#M1481 I'm trying to fit a generalized additive model as follows:<BR /> <BR /> proc gam data=dataset; <BR /> class DOW season holiday;<BR /> model deaths=param(DOW season holiday) <BR /> spline(date) <BR /> spline(temperature)<BR /> spline(humidity)<BR /> spline(NO2)<BR /> spline(Ozone)<BR /> spline(SO2)<BR /> spline(PM25)<BR /> spline(PM10) / dist=Poisson method=gcv nodev=norefit; <BR /> output out=dataset_out p;<BR /> run;<BR /> <BR /> However, I keep getting an error message that states, "The backfitting algorithm diverged for this model."<BR /> <BR /> What does this mean and how can I fix it? Thanks! Thu, 21 May 2009 20:36:43 GMT deleted_user 2009-05-21T20:36:43Z https://communities.sas.com/t5/Statistical-Procedures/Proc-GAM/m-p/35376#M1481 I'm trying to fit a generalized additive model as follows:<BR /> <BR /> proc gam data=dataset; <BR /> class DOW season holiday;<BR /> model deaths=param(DOW season holiday) <BR /> spline(date) <BR /> spline(temperature)<BR /> spline(humidity)<BR /> spline(NO2)<BR /> spline(Ozone)<BR /> spline(SO2)<BR /> spline(PM25)<BR /> spline(PM10) / dist=Poisson method=gcv nodev=norefit; <BR /> output out=dataset_out p;<BR /> run;<BR /> <BR /> However, I keep getting an error message that states, "The backfitting algorithm diverged for this model."<BR /> <BR /> What does this mean and how can I fix it? Thanks! Thu, 21 May 2009 20:36:43 GMT https://communities.sas.com/t5/Statistical-Procedures/Proc-GAM/m-p/35376#M1481 deleted_user 2009-05-21T20:36:43Z https://communities.sas.com/t5/Statistical-Procedures/Proc-GAM/m-p/35377#M1482 My hunch is that, Proc GAM may not be able to fit the non-linear curve. Try simplying the model by including less predictor variables..not sure if this will work Fri, 22 May 2009 13:55:57 GMT https://communities.sas.com/t5/Statistical-Procedures/Proc-GAM/m-p/35377#M1482 deleted_user 2009-05-22T13:55:57Z