Hi, there,
I am modeling some high degree polynomial terms in my regression model, the output looks like:
Parameter | DF | Estimate | Error | t Value |
Intercept | 1 | 0.813558 | 0.147240 | 5.53 |
A | 1 | 0.916600 | 0.030994 | 29.57 |
B | 1 | 0.257152 | 0.094209 | 2.73 |
C*D*E_ | 1 | 0.023605 | 0.004686 | 5.04 |
B^3*C* | 1 | -0.000356 | 0.000105 | -3.40 |
My polynomial degree was set to 5, and it really looks weird. Obviously the last two parameters are missing some terms. How can I fix it? Any comment is highly appreciated.
Hi Viva,
More info please.
Are you using Enterprise Miner? The maximum polynomial degree is 3 for a regression node. How are you setting it to 5?
Thanks,
-Miguel
Thanks for clarifying which proc you are using.
If you look closely at the log of your run, you notice that by default PROC glmselect has stepwise selection turned on. That is why you will only see polynomial terms if they were kept after the selection.
I googled PROC GLMSELECT documentation for you. In case you want to turn off the selection add the option " / selection=none" in your proc statement.
e.g.
model amount = MyPoly / selection=none;
This will make your polynomial terms appear there as you expect, although a selection method is highly recommended!
Good luck!
Miguel, thanks for your suggestion.
I do need to use stepwise method to make sure my model is only including those significant factors. What I am saying here is about the "output display" error.
Let us look at the last term B^3*C*, I highly suspect the star symbol after C, it should look like: B^3*C*(something missing here?, maybe A or E) This have nothing to do with the selection method here, I am doubting the output is incomplete.
This happened in my example as well, and I fixed it specifying a larger length with the namelen option.
E.g.
proc glmselect DATA=sampsio.dmagecr namelen=50;
effect MyPoly = polynomial(duration checking savings/degree=4);
model amount = MyPoly / selection=none;
run;
Curious why you would have an issue if your variables are so short. Please give this a try and let me know how it went.
Thanks,
Miguel
Are you ready for the spotlight? We're accepting content ideas for SAS Innovate 2025 to be held May 6-9 in Orlando, FL. The call is open until September 25. Read more here about why you should contribute and what is in it for you!
Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.
Find more tutorials on the SAS Users YouTube channel.