About bncoxuk

Pritish · ‎11-21-2011

Hi, Even it worked for me! Here's what I have: data data01; input key col1; cards; 1 101 2 102 3 103 4 104 5 105 ; run; data data02; input key col1; cards; 1 101 <-- This col is same as previous dataset 6 102 7 103 8 104 9 105 ; run; data want; merge data01 data02; by key; run; Result Set: Key Col1 1 101 2 102 3 103 4 104 5 105 6 102 7 103 8 104 9 105 Do you have the same column name in all the tables? If possible provide example of what you have.

Howles · ‎11-18-2011

The ARRAY isn't necessary. Instead: length newv $ 200 ; newv=catx(' ',of _numeric_) ; art297 wrote: data have; do i=10 to 11; n1=i; n2=i+8; n3=i+5; n4=i+1; n5=2*i; output; end; run; data want; set have; array nums(*) _numeric_; newv=catx(' ',of nums(*)); run;

SteveDenham · ‎11-17-2011

Quite possibly a mixture problem. It looks like 0.13% of the data significantly elevate the intercept. That can happen. Consider the mean net worth of a very rural county, where 200 people live, say $50,000 per person. All of a sudden Bill Gates moves in, worth say $20B. The mean net worth is now 1991 times as large, with only a change of 0.5% of the data. I would wager that those 1307 exceptional cases, when examined separately, tell you something quite interesting. Steve Denham

bncoxuk · ‎11-16-2011

Thanks for your suggestion, art297. The variables are not 100% perfectly correlated. Just a level of them. So I cannot remove variables. I found a way to deal with that: 1) recode the levels of variables into 0, -1, 1 (effect coding); 2) Only keep one of the correlated levels in the final model.

GreggB · ‎11-08-2011

I use something like this. If you use the RETAIN statement it will force the vars to stay in the order you want them. Notice the RETAIN statement comes before the SET statement. data final; retain schoolid school Student_Num Last First Middle Grade Engprof Ethnicity Sex ; set joinall; school=put(schoolid,psfmt.); keep school EngTch Engprof Ethnicity First GTclass GTmark GTtch Grade IEP LangRIT Last MathRIT MathTch Middle ReadRIT SciTch Sex ; run; proc export data=final outfile='G:\Departments\Research\MAP\1112\MAP - 1stQ_Grades - PASS\Fall MAP 1stQ Grades PASS - school..xls' replace; run;

art297 · ‎11-07-2011

DN, Totally agree! But, I think it is really just a matter of preference as there is virtually NO performance difference between the two methods (at least when tested on 1.9 million records).

bncoxuk · ‎10-18-2011

I found a solution: just use PROC LOGISTIC, which can estimate type3 for sex and married when inter is specified (rather than sex*married).

bncoxuk · ‎09-20-2011

Hi art, your code is also correct. As Tom said, I should have put capital letters in the libname and memname statement. Thank you very much.

ballardw · ‎09-26-2011

Peter's solution is better than mine in that you won't reference any non-existant data sets. Mine was just to show one approach using an explicitly iterated %DO loop using the obvious pattern shown in the data set names.

art297 · ‎09-13-2011

You apparently didn't notice the title of that particular exhibit in the paper it came from: "Figure 6. A Common Error" see: http://www.nesug.org/proceedings/nesug03/bt/bt009.pdf In short, the code won't work as written. However, it is also a nice summary on the topic.

trekvana · ‎08-23-2011

I tend to agree with DLing here. Usually the issue lies with as DLing said - poorly gathered and/or irrelevant data. You need to ask yourself if the data collected is pertenent to the hypothesis you are testing. If so then maybe the poor fitting model is a sign that the variables you collected are not influencial to your dependent variable which could be a finding in of itself. There have been many times where I have seen clinical protocols that try to ask questions to with which the data collected cannot even answer.

Rick_SAS · ‎08-23-2011

If you get a "convergence is questionable" warning, then I would not use that result. Although both GENMOD and LOGISTIC estimate parameters by maximizing a log-likelihood, they might not use the same optimization method or starting conditions, so one could converge and the other not. The LOGISTIC doc says "the maximum likelihood estimation is carried out with either the Fisher scoring algorithm or the Newton-Raphson algorithm" and the default technique is Fisher scoring, which is an iteratively reweighted least squares algorithm. For details, see http://support.sas.com/documentation/cdl/en/statug/63962/HTML/default/viewer.htm#statug_logistic_sect033.htm

Reeza · ‎08-19-2011

For regression with categorical variables essentially dummy variables get created for n-1 levels with the nth level being all 0. So if one level correlates perfectly with another level in a different categorical variable, the two are linear combinations of each other, same as Doc@Duke example. Typically you can change the way you categorize the variables to help solve this one, but in your case with all missing for a particular population it might be tricky. You also have systematic missing data, which is a problem in of itself.

Doc_Duke · ‎08-17-2011

Your data are overparameterized. See my response in http://communities.sas.com/thread/30685?tstart=0 Doc Muhlbaier Duke

bncoxuk · ‎08-17-2011

Yeah, it is pretty hard to describle. I will try to think for a while. Thanks.

Online Status	Offline
Date Last Visited	‎09-01-2015 07:11 AM

Re: How to concatenate character values across variables

How to concatenate character values across variables

How to concatenate character values across variables

How to create a report by combining results from simple PROC SUMMARY

LIKE '%' does not work

What is the functionality of || here

What is the functionality of || here

What is the functionality of || here

Re: How to get the percentage for each unique combinations of variable...

How to get the percentage for each unique combinations of variables

How to set all missing values to zero for all variables?

How to merge files with similar names in the same library

Re: How to convert all numeric variables into characters and concatena...

Overall mean of general linear model

Categorical variables have an identical level

Output data to Excel

How to convert numeric values to characters without adding blanks

Interaction effect in PROC GENMOD

How to combine files with a macro

How to combine files with set

A simple sas macro

Poor model fit with binary logistic regression

Some interesting results: PROC GENMOD vs PROC LOGISTIC

The effect of one variable's level cannot be estimated if...

Please look at a simple data and a tricky question

Why the coefficients are 0