Hello,   You should be able to use an aggregated measure in a line chart. What happens when you try?   If the aggregated measure is based on a periodic operator such as CumulativePeriod, it will return missing values if the chart does not contain the same date data item that the periodic expression is based on.   Sam ... View more

Hello,   I believe you could calculate this by using two of the periodic operators in VA:   CumulativePeriod will return the year-to-date value for each quarter if you specify ByQuarter as the inner interval and ByYear as the outer interval. Period will return the value for the current year if you specify ByYear as the interval.   So, you could make an expression like this: CumulativePeriod(_Sum_, 'Expenses'n, _IgnoreAllTimeFrameFilters_, 'Quarter'n, _ByQuarter_, _ByYear_, 0, 1, _Full_, {Date}) / Period(_Sum_, 'Expenses'n, _IgnoreAllTimeFrameFilters_, 'Quarter'n, _ByYear_) And get a result like this:   This is not exactly the same thing as the data in your opening post. You need the quarter and year to be the same column for this to work. In addition, your report object can only have one date column it for the periodic functions to work. You could work around this somewhat by making an extra character column with the years, for example.   Let us know if that helps, Sam ... View more

There are a few different ways to control which variables are used in the model. 1) on the Data tab you can set the variable roles (as input, rejected (not used), or other roles). These will get used as the defaults for all models in all pipelines in your project. 2) Within a pipeline, use the Manage Variables node (as you are) to set the inputs for any models that follow this downstream 3) Use the Variable Selection node to intelligently automatically determine which variables are most influential (with a number of techniques for this available)...as opposed to manually selecting them yourself. https://go.documentation.sas.com/doc/en/capcdc/v_018/vdmmlcdc/vdmmlref/p1l7tl7hddl0lon138uretl5muac.htm   You can see hat inputs were used in the model by opening the Results for the modeling node after it was run. Hope this helps. ... View more

@znhnm wrote: Exactly! The calculation is correct! Thanks a lot! However, can I add tis pctn column in my dataset as the 3rd column? I also have anoter column called number_of_items and I want to calculate the pctn2 as the 4th colummn in my current dataset Now you have to provide actual example data. Typically this sound like calculate and then recombine with your existing data. Proc Tabulate will create output data sets but it takes a bit to get used to the content. Add an OUT=<datasetname of choice> to the Proc tabulate statement and you can then look at what one looks like. My example would create a pretty simple data set but there will will be variables you don't want related to more complex output possible with the procedure. If your other data isn't too ugly it should be possible to MERGE with a data step or join using proc sql.   ... View more

It means that some (actually most) observations can't be used because they contain missing values. You might want to consider imputing values for these missings.   When you apply a filter (you didn't say what filter, so I am guessing), there are fewer observations that can be used. ... View more

Hi, Currenlty the only way to consume the attribution results is within the interface due to the complexity of the models and data presented.   Hope this helps! Jeffrey ... View more

I personally refuse to work with data in such a horrible (spreadsheet) format. I recommend you do that too. Apparently, @ballardw also recommends you do that.   I would transpose the data so that calendar information is now contained in a variable.   data a; do i = 1 to 10; reference_month = int(12*ranuni(123)); Jul21 = int(100*ranuni(123)); Aug21 = int(100*ranuni(123)); Sep21 = int(100*ranuni(123)); Oct21 = int(100*ranuni(123)); Nov21 = int(100*ranuni(123)); Dec21 = int(100*ranuni(123)); Jan22 = int(100*ranuni(123)); Feb22 = int(100*ranuni(123)); Mar22 = int(100*ranuni(123)); May22 = int(100*ranuni(123)); Apr22 = int(100*ranuni(123)); Jun22 = int(100*ranuni(123)); Jul22 = int(100*ranuni(123)); Aug22 = int(100*ranuni(123)); Sep22 = int(100*ranuni(123)); Oct22 = int(100*ranuni(123)); Nov22 = int(100*ranuni(123)); Dec22 = int(100*ranuni(123)); output; end; run; proc transpose data=a out=b prefix=value; by i reference_month; var jul21--dec22; run; data c; set b; which_month=input(_name_,monyy.); prev_col1=lag(value1); prev_col2=lag2(value1); prev_col3=lag3(value1); prev_col4=lag4(value1); prev_col5=lag5(value1); prev_col6=lag6(value1); if month(which_month)=reference_month then do; score0=value1; score1=prev_col1; score2=prev_col2; score3=prev_col3; score4=prev_col4; score5=prev_col5; score6=prev_col6; end; if n(of score:)>0 then output; format which_month monyy5.; drop jul21--dec22 _name_ prev_:; run;   Since your original data has Apr22 and May 22 in the wrong order, I haven't tried to fix that at all, but then of course the results have wrong ordering. I leave it as an assignment for you to fix.   Also, @znhnm , please help us (which helps you get faster and better answers) by always providing example data, as working SAS data step code. Don't wait for us to ask, just do it. Thank you. ... View more

Thanks lot for this! I was trying to find a solution to have it on both number on 3-letter format. I have it both with your way which is amazing. ... View more

Please post the complete(!!!) log of both the step where you create the numeric variable and the PROC SORT. Use this button to pist the log text: ... View more

Don't use PROC IMPORT, write the DATA step yourself. IMPORT is not needed for text files and causes unnecessary problems. See thousands of posts here.   As a stopgap measure, set options validvarname=v7; before PROC IMPORT. ... View more

I am assuming you are looking at some PRINT OUT of the values that makes you think that string is there.   The usual cause is leading spaces.  Especially if the print out you looked at was generated with ODS as that REMOVES the leading spaces when it generates the print out. if left(reason) = "asd asd asd" then ... Other obvious things to check.  Does the CASE of the letters agree?  ASD is a different string than asd. if lowcase(left(reason)) = "asd asd asd" then ... And the harder problem is non-printing characters in the string.  So they look like spaces, but they aren't spaces. Things like tab ('09'x) or linefeed ('0A'x) or carriage return ('0D'x) or non-breaking spaces ('A0'x) . if lowcase(left(translate(reason,' ','090A0DA0'x))) = "asd asd asd" then ... or perhaps something more exotic.  To look for those print the values using the $HEX format. ... View more

Hi Koen, I realized that I made a typo. Sorry about it. Your suggestion works perfectly. Appreciated your help! ... View more

Quite often  a custom format is a better way than creating new variables. A group created by formatting values of a single variable will be honored by reporting and analysis procedures and generally will work with graphing.   Here is an example of creating a custom variable for an age variable in a data set you can test code against. Proc format; value agegroup low-12 = 'Preteen' 13- 18 = 'Teen' 19 - high= 'Adult' ; proc freq data=sashelp.class; tables age; format age agegroup.; run; The data set Sashelp.class does not have any ages over 16 so you don't see them in the result.   Advantages of a custom format are: 1) You don't need to add variables. In large data sets such may take a noticeable amount of time. 2) If you want to change the group then just modifying or creating another format is much faster than adding yet another variable. 3) If you want to apply the same logic to multiple variables (such as age at enrollment, age at graduation) then you can just by associating those variables with the format. 4) The logic for ranges of values in formats is often easier to write than the possibly horrific number of "if/then" statements that might be needed to create a variable. The character < is used in value list to indicate whether the end value is excluded or not of a range. 5) if the variable is numeric then the order of display will generally follow the underlying numeric values. If you create character variables then quite often the default order of appearance in results will be alphabetic.   Note: some caveats. If you variable(s) are character the format name on the value statement in proc format needs to include a $ and used with a $. Character values should be included in quotes. Multiple specific values, whether numeric or character are separated by commas. You cannot assign the same value to different results. There are some options, such as OTHER available to say "any value not specifically listed is displayed as" the shown value.     ... View more

@LinusH wrote: I don't have access to SAS Viya at this point, but I was assuming (maybe incorrectly) that if you have a libref assigned to your caslib, that it contents would show up in the PROC SQL DICTIONARY...? I also believe it would show the tables but not necessarily a row count - similar to what you'd see if it would be a libref pointing to a database table. ... View more