thank you, this worked for me, so i was doing %IF instead of IF and that caused the issue. ... View more

Good. Hope one day I could come to London and see the Big Ben . ... View more

Transposing the values it not really that hard IF you understand your data. Let's assume that you have some set of key variables that identify the group of values that go together.  Then make sure the data is sorted by those and process each group to create the three variables.   Since your existing dataset is too unwieldy to serve as a useful explanatory example code let's make up some clearer variable names.  Say how a vertically structured dataset that consists of a group of subjects that each have one or more visits and got one or more multipart questions.  So we might use variables named SUBJECT, VISIT, QUESTION, SUBQUESTION, VALUE to represent that data.  Your problem is that it kind of looks like you mgiht have QUESTION and SUBQUESTION mashed up into one variable?  But perhaps one of those other variables with the codes can be used to group the questions together?  If not then add a step to tease out the question and subquestion from the text.   So now to reduce the dataset to one observation per question that has answers to the three subquestions as separate variables you might do something like:     data wider ; do until(last.question); set have; by subject visit quesiton ; if subquestion='SEVERITY' then SEVERITY=value; else if subquestion='FREQUENCY' then FREQUENCY=value; else if subquesiton='INTERFERENCE' then INTERFERENCE=value;      end;    drop subquesiton value; run; Now you can sort D1 and WIDER by SEVERITY FREQUENCY INTERFERENCE and merge them to pull any lookup values from D1 and attach them to WIDER.   If you need to revert the result back the TALL structure then you will need to reverse the process, perhaps needing the define new values of the QUESTION/SUBQUESITON pair to be able to store the value of the variables pulled from D1 into the TALL structure.   ... View more

Well, you can have SAS generate and run the VBS if really necessary. ... View more

Thanks for the reply Tom, sorry to say that proc factor and proc score did not helped me.   What actually i am looking for is there are categories in the scoring dataset called Frequency severity and interference, they have a specific scoring combination based on paramcd, FSI means this paramcd is present in all three categories and  some other paramcd's are present in only one or two categories.  for example for paramcd PRO0117 in the first, second and third row the values are as Never,0,None,0,Not at all,0,0,PRO0117,FSI Rarely,1,None,0,Not at all,0,0,PRO0117,FSI, Rarely,1,None,0,A little bit,1,1,PRO0117,FSI, In the first row if all 3 categories are 0 then the composite is 0 and in the second row if Frequency is 1 and severity is o and Interference is 0 even then the composite is 0 and in the third row if the frequency is 1 and severity is 0 and interference is 1 then the composite is 1. The similar pattern of composite scoring is present in dataset b and based on the dataset a we need to map it on dataset b to get the accurate composite score.   So this dataset is a reference for us and i had not merged this to dataset b and instead of that i had mapped everything manually to get the composite score and calculate the further aval value analysis.   Here i was requesting if there any solution where we can avoid this manual process and map them with the dataset b. In the dataset a we have got scores represented in the horizontal format as a  reference and this will be applied to any subject, but in dataset b the scoring variable is present in vertical format, so do you have a solution to do this automatically.   I hope i did not confused you, please let me know if you need any clarification from my end. thank you   ... View more

Without data there is no way to tell what you may be "doing wrong". Either provide some example data in the form of data step code or rewrite your code to use a data set in the SASHELP library.   A typical approach to determining which of your options are not playing well with your graph is to start with very sparse code. Then add one feature/option at a time. For instance start with proc sgplot data=total_dup ; series x=d_VISITN y=total / group=d_SUBJID ; xaxis label="Visit" values=(1 to 4 by 1); yaxis label="Total" values=(0 to 5 by 1); format d_VISITN visit. domain_score $score.; run; Then start adding ONE option at a time.   If you only have one color in the output then there are some possible issues related to data such as only one value of your group variable has values for the x and y variables in the ranges supplied on the XAXIS and YAXIS values= options. Another possible is there is only one value for the group=variable. Which is why we ask for example data.   Note that since the Styleattrs(datacontrastcolors)= option and the Grouplc= option are attempting to set colors for the same objects, the lines, when they conflict the results are unpredictable and especially without any actual data. ... View more

Might it not be better overall to transpose to a long format and sort with nodupkey? That would eliminate ALL duplicate values. ... View more