When I add more independent vars, without changing any options, I'm seeing that the resulting model can be worse than with fewer independent vars. The fit statistics like misclassifications and AUC are worse. Why is that?    I would think if the new vars don't contribute, it would just get ignored, not make the model worse. This makes me wonder what my strategy should be for adding more vars. My understanding is that an advantage of decision trees is that I can essentially throw in the kitchen sink and HPSPLIT will figure out what to use and what not to use. Naive of me?   ... View more

I need to express my linear regression model as an equation so that I can calculate the predictions separately (ie, outside SAS). Ditto for my logistic regression model. Could you confirm whether I'm doing this correctly?   --------------------------------------------------------------------------------- LINEAR regression parameter estimates (using proc genmod) Intercept: 3.9273 var1: 0.0015 I've expressed the equation as:     prediction   =   3.9273 + 0.0015*var1   --------------------------------------------------------------------------------- LOGISTIC regression parameter estimates (using proc logistic, binary, solving for "1") Intercept: -2.8240 var2: 1.2638 I've expressed the equation as:    probability   =   exp(-2.8240 + 1.2638*var2)   /   (1 + exp(-2.8240 + 1.2638*var2))   --------------------------------------------------------------------------------- Yes, no, maybe so??? I need to confirm because when I do this, the answers do not match what I get when scoring in SAS directly. Thank you!  ... View more

I have a small dataset of x and y values that create a downward trending curve. Using my vast statistical vocabulary, it looks like an "L". I'm trying to determine the equation for it. Looks like I can use proc transreg to fit the curve but how do I output the equation? ... View more

The current code... proc univariate data = my_input_dataset; var age weight height; output out = my_output_dataset median = median mean = mean mode = mode ; run; So this appears to output the stats for "age", but then overwrites with the stats for "weight", and finally "height". So I only end up with one row containing the stats for the most recent variable. Can the univariate procedure create a dataset with the stats for all of these variables (ie, without having to throw in some data steps and loop, etc)? In which case, is there a way to throw in the variable name to distinguish the rows? ... View more

Can't figure out how to address this warning... WARNING: The variable l12_drop_giv_obchan_other has zero or negative sum of squares after being adjusted for the intercept, if there is one. Hence, it will not be used in any subset selection. Then the procedure stops. My code is this... proc reg data = input_dataset; model dep_var = <<< 32 independent vars >>> / selection = rsquare adjrsq cp best = 3 noint ; run; Funny thing is it runs fine when there are 31 variables, but not when I add more. And it doesn't seem to matter which variable I added. All the values in the variables appear to be fine. I can't seem to find any documentation about this warning. Anyone have some ideas? Thank you. ... View more

This is more a general regression question than specific to sas procs... Let's say I'm trying to create a model that predicts how much my customers are likely to spend at my store. But looking at my existing customers, I have a hunch that customers who have bought product X are different than those who have bought product Y. In other words, if I were to build a model from product X customers, the final variables and coefficients could be very different from the model built from product Y customers. Do I have to build a separate model for each customer segment and somehow blend the results? Or is there a way to build one model that somehow factors in the differences? ... View more

Seems pretty simple but can't seem to find documentation about this... How do I reverse the order of the values so the odds ratio is the inverse/converse/whatever(!)? For instance, if the values of the variable of interest were 'female' and 'male', the odds ratio would default to male over female... but I'd like to get female over male. ... View more

I have about 20 numeric variables that can potentially have a null value. I'd like to convert all of them into zero. I don't want to list out each variable individually. I know I've done this before with just a few lines of code, maybe with an array, but can't seem to remember exactly. Suggestions? Thanks! ... View more