topic Non contributing variable in multiple linear regression in Statistical Procedures

Non contributing variable in multiple linear regression

heydaloran — Fri, 01 Sep 2017 22:11:22 GMT

Hi! I'm running a multiple regression procedure, and we were asked to "obtain a 95% interval estimate of the mean rental rate for the following specifications of the explanatory variables: age_of_prop=5, operate_exp=8.5, vacancy_rate=0.15, total_sqft=250000. My issue is that when I create a new data set to obtain this specific information, the dataset returns 0 values, so nothing prints.

Is my "AND" boolean logic correct in the IF statement, or should I be using something different to say "this and that and that"?

Here I've just set up the regression procedure - it outputs to the set "results"

proc reg data=cre;
model rate_per_sq= age_of_prop operate_exp vacancy_rate total_sqft
/stb clm cli clb alpha = .01;

output out=results predicted=yhat LCLM=LCLMEAN UCLM=UCLMEAN LCL=LCLPRED UCL=UCLPRED STDP=SEmean STDI=SEi;

run;

Here is where I establish the new dataset. The if statement works if it states something obvious like "IF(1=1 and 2=2)," but I can't get it to return data when I want it to predict the rate_per_sq when the statement is defined as it is below.
data results_1;
set results;
if(age_of_prop=5 AND operate_exp=8.5 AND vacancy_rate=0.15 AND total_sqft=250000);

run;

proc print data=results_1;
run;

Re: Non contributing variable in multiple linear regression

ballardw — Fri, 01 Sep 2017 23:22:50 GMT

Make sure that your input data set has the combination of values that you want to evaluate but make sure that the RATE is missing. I have to assume the your "rental rate" is the rate_per_sq variable since you do not state so explicitly (hint: indicate which variable you mean).

You will find the predicted value and the associate confidence limits in your results output set.

Re: Non contributing variable in multiple linear regression

heydaloran — Sat, 02 Sep 2017 19:33:58 GMT

You're right! I forgot that the IF statement can't draw predictive conclusions. it simply looks for and selects the values in the dataset, of which they didn't exist..

What I ended up doing is appending the data it requested to the original dataset. Then I ran the same statement and it worked fine.

Thanks for your help!