qa_package_pd

/****************************************************************************************

*                                         SENTINEL PROGRAM

*****************************************************************************************

* NAME:             scdm_data_qa_review-maskresults.sas

*

* PURPOSE:          This program evaluates select QA package output based on user-defined parameter

*                   to identify whether conditions for masking apply (results specific to select TabID)

*                   and numeric value between 1 and user defined max inclusive. If conditions met,

*                   the numeric value will be masked alongside other related statistics to prevent back

*                   calculation

* 

* MAJOR STEPS: 

*                   1) Evaluate QA package output to identify files where masking may be indicated

*                   2) Evaluate each file to define conditions for and variables to mask

*                   3) Evaluate masking conditions and if met, mask numeric variable values

*                   4) Evaluate count of masking updates made per file. If only one, mask additional records in file

*                      per conditions previously defined to prevent back calculation

*

* KEY DEPENDENCIES/CONSTRAINTS/CAVEATS:

*

*   PROGRAM INPUT:  QA Package MSOC folder output   

*

*   PROGRAM OUTPUT: Masked QA Package Output (note file output to MSOC directory will be 

*                   evaluated for masking but only updated to masked if conditions met to do so  

*

****************************************************************************************/

​

​

    /* Create temp subdirectories in dplocal for organizing masking process */

    options DLCREATEDIR;    

            libname mmsoc "&MSOC.mask";

    options NODLCREATEDIR;

​

    /* create concatenated libname to reference data for masking, selecting mask version if it exists rather than original version from msoc */

    libname maskin (mmsoc msoc) access=readonly; /* concatenated libref to mask multiple tabids - if masked version exists, this version will be selected when macro iterates otherwise dataset from qapackage msoc will be selected*/

    %let maskMin = 10;/* max value for matching - masking will be triggered if value between 1 and this value inclusive */ 

​

​

    /* known numeric variables that should not be used to determine masking requirement but masked if requirement met by other variable within row of results */

    %LET EXACTMATCH_STR = %STR("MIN" "P1" "P5" "P25" "MEDIAN" "P75" "P95" "P99" "MAX" "MODE", "PROPORTION_FLAGGED"); /* if exact match on variable name, space delimited quoted list*/

​

​

*** process data to apply matching                                                                 ***;

​

    /* Select list of datasets and variable where masking could apply i.e. not signature or metadata  */

    Proc sql NOPRINT ;  

        create table work._DS_SELECT as 

        select distinct   upcase(memname) as memname

                        , upcase(name) as name

                        , upcase(type) as type

                        , length 

        from dictionary.columns 

        where      libname='MSOC'

               and memtype = 'DATA' 

               and memname not in ('ALL_L1_SCDM_COMP', 'ALL_L1_CONT', 'DTH_MODULE_ALL_L1_CONT')

        order by memname, name;

    quit;

​

​

    *** Macro to define and apply masking conditions per qa dataset                                    ***;

    %macro filter_memlist (tabid_list = );

​