Hi,   I'd like to have your help with counting unique individuals per usage (code in my case) before I spend too much time on working out nothing. Basically I tried to count unique individuals per diagnosis/procedure code. An individual can have multiple records, and in each record there are 10 diagnosis codes and 5 procedure codes. Many codes are actually invalid (having values like 000000 and 999999). There are nearly 50 eligible diagnosis codes and 30 eligible procedure codes to check. I attached some sample data for more details.   Could anyone guide me how to count unique individuals per eligible code?   Thanks a lot, Lizi ... View more

Hi all,   I am using the INFLUENCE option to get diagnostic plots for logistic models. Is it possible to suppress influence the diagnostic output but keep the plots?   Thanks a lot, Lizi ... View more

Hi Everyone,   I'm trying to do two-way ANOVA check with Dunnett option but need to have p-value for the mean comparison. I use the code below:   proc anova data=dataset; class A B; model COST = A | B;  means A | B /dunnett;  run;   The last piece of the test has the following table:   I need to have P-values to see if means are significantly different across the groups. Anyone knows how to do it?   Thanks a lot! Lizi   ... View more

Hi everyone,   I'd like to get help with generating a similar plot as below. The data has the continuous variable COST and two categorical variables A and B. I wanted to compare COST of different groups of A and B through the visualization.    Anyone has a suggestion about where to start? PROC ANOVA, PROC SGPLOT or PLOT GLM?     Thanks a lot! ... View more

Hi everyone,   I am trying to do ttest of two groups to see if the group means are different from a fixed value. How can I specify the null value in PROC MEANS ttest?   proc means data=mydata t probt;    class flag;    var cost; run;   Thank you! Lizi ... View more

Hi everyone,   This is a similar question to what I posted last time.   I tried plug in a date macro variable (&curdat_2mon as in the code below) in my query. But it looks the quote (%str(%')) I used is not right? Can anyone help me correct again? If there is a simplified way to use the macro variable in the query, that would be great too. Thanks a lot!   data _null_; curdat_2m = intnx('Month',"&sysdate"d,-2,'s'); call symput ('curdat_2mon',put(curdat_2m,date9.)) ; run; %put &curdat_2mon ; PROC SQL; CREATE TABLE new_tbl as select distinct MBR_ID, FST_NM, LST_NM, DOB, ZIP from DB.MBR_TBL where  %str(%')&curdat_2mon.%str(%')d between BEG_DT and END_DT ; QUIT; ... View more

Hi everyone,   I am trying to plug in a date macro variable (&curdat_2m as in the code below) in my query, but somehow I got no rows. When I used specific date like 20070515, the query ran fine. Can you see what problem with the where clause in the query? Thanks a lot!   data _null_; curdat_2m = intnx('Month',"&sysdate"d,-2,'s'); call symput ('curdat_2m',put(curdat_2m,yymmddn8.)) ; run; PROC SQL; CONNECT TO ORACLE( ...); CREATE TABLE new_tbl as SELECT * FROM CONNECTION TO ORACLE ( select distinct MBR_ID, FST_NM, LST_NM, DOB, ZIP from DB.MBR_TBL where '&curdat_2m.' /*if using '20070515', the query generates a number of rows*/ between to_char(BEG_DT, 'YYYYMMDD') and to_char(END_DT, 'YYYYMMDD') ); DISCONNECT FROM ORACLE; QUIT;   Thanks for the help! Lizi ... View more

Hi everyone,   I need to productionalize data formatting instead of working on an excel file and are getting some questions. I need to add a row to sum up the counts by group and keep the row on the top of each group. Specifically, below is the sample data I have:   questionid question answer count 1 Age 0-64 29 1 Age 65-74 766 1 Age 75-84 161 1 Age 85+ 25 2 What's your marital status and current living situation?: Marital Status DIVORCED 63 2 What's your marital status and current living situation?: Marital Status DOMESTIC PARTNER 14 2 What's your marital status and current living situation?: Marital Status MARRIED 393 2 What's your marital status and current living situation?: Marital Status SEPARATED 3 2 What's your marital status and current living situation?: Marital Status SINGLE 42 2 What's your marital status and current living situation?: Marital Status WIDOWED 79 3 What's the annual gross income for your household? $0.00 - $18,000 29 3 What's the annual gross income for your household? $18,001 - $50,000 194 3 What's the annual gross income for your household? $50,001 OR MORE 338   I would like to have the format as below: questionid question answer count percent 1 Age TOTAL 981 1 1 Age 0-64 29 0.029562 1 Age 65-74 766 0.780836 1 Age 75-84 161 0.164118 1 Age 85+ 25 0.025484 2 What's your marital status and current living situation?: Marital Status TOTAL 594 1 2 What's your marital status and current living situation?: Marital Status DIVORCED 63 0.106061 2 What's your marital status and current living situation?: Marital Status DOMESTIC PARTNER 14 0.023569 2 What's your marital status and current living situation?: Marital Status MARRIED 393 0.661616 2 What's your marital status and current living situation?: Marital Status SEPARATED 3 0.005051 2 What's your marital status and current living situation?: Marital Status SINGLE 42 0.070707 2 What's your marital status and current living situation?: Marital Status WIDOWED 79 0.132997 3 What's the annual gross income for your household? TOTAL 561 1 3 What's the annual gross income for your household? $0.00 - $18,000 29 0.051693 3 What's the annual gross income for your household? $18,001 - $50,000 194 0.345811 3 What's the annual gross income for your household? $50,001 OR MORE 338 0.602496   Can anyone help me out?   Thanks a lot! Lizi ... View more

Hi All,   I tried to get the lower triangle of a correlation matrix with the code below. When I used the variables and specific number of variables (do i=1 to 3 instead of do i=1 to dim(pval) ), the coding worked fine. When I tried to use macro variable and dim() function as below, the output was misreplaced (having wrong correlation values except the first two columns). Could anyone help me correct my code?   %let XVARS= var1 var2 var3;   **Get full correlation matrix; ods output PearsonCorr=fullcorr; proc corr data=mdl; var &XVARS.; run;     **Keep statistically significant coefficients; data corr; set fullcorr; array pval{*} p&XVARS.; array vrbl{*} &XVARS.;        do i=1 to dim(pval); if pval[i]>=0.05 then vrbl[i]=.;      end; run;   data corr; set corr (keep=variable label &XVARS.); run;   **Get lower triangle matrix; data lower_tri; set corr; array vrbl{*} &XVARS.;    do _i=1 to dim(vrbl); if (_i>_n_) then vrbl[_i]=.;      end;      drop _i; run;   Thanks as always! Lizi ... View more