Base SAS: Transposing Counts By Date Var

Question

I need to create a summary dataset/report which tracks the flow of these purchases over time.I have a dataset which gives a signup date for an overall service and 9 variables which give the purchase dates for different add on products. If the add on variable dates match the signup date then those add on products were included with the signup package. Any add on variable purchase date that comes after the signup date are products which are purchased during the history of the active account. This is what it looks like:

data have ;
 length ID 8 
  signup_DT 8 preferredhd_tv_estbd_dt 8 
  ultimate_estbd_dt 8   quant_estbd_dt 8    
  FullyLoaded_estbd_dt 8    HB_estbd_dt Cin_estbd_dt 8  
  time_estbd_dt 8   router_estbd_dt internet_estbd_dt 8; 
 INPUT ID 8 
  signup_DT  : anydtdte9.  preferredhd_tv_estbd_dt  : anydtdte9.    
  ultimate_estbd_dt  : anydtdte9.   quant_estbd_dt  : anydtdte9.    
  FullyLoaded_estbd_dt  : anydtdte9.    HB_estbd_dt Cin_estbd_dt  : anydtdte9.  
  time_estbd_dt  : anydtdte9.   router_estbd_dt internet_estbd_dt  : anydtdte9. ;;
 format   signup_DT  preferredhd_tv_estbd_dt    
  ultimate_estbd_dt     quant_estbd_dt  
  FullyLoaded_estbd_dt  HB_estbd_dt Cin_estbd_dt    
  time_estbd_dt     router_estbd_dt internet_estbd_dt   date9.; 
datalines; 
98663699    4/7/14  4/9/14  4/7/14  9/12/14 10/15/14 7/7/14 4/7/14  4/7/14  4/12/14 .
33663798    4/11/14 .   4/11/14 .   4/11/14 4/11/14 4/11/14 4/11/14 6/11/14 7/15/14
43663463    5/12/14 5/12/14 5/12/14 9/5/14  9/17/14 .   .   .   .   .
77661437    5/16/14 .   5/16/14 .   10/31/14    .   5/16/14 5/16/14 11/16/14    .
85662295    5/29/14 .   .   5/29/14 .   6/12/14 .   .   11/16/14    .
36656756    6/4/14  .   .   .   6/4/14  6/4/14  6/12/14 6/4/14  6/4/14  12/4/14
67662646    6/14/14 .   6/14/14 8/31/14 .   .   6/17/14 6/14/14 .   6/22/14
55663786    6/26/14 .   .   .   8/14/14 6/26/14 7/8/14  6/26/14 11/30/14    .
44663191    8/21/14 .   9/30/14 .   .   .   .   1/12/15 .   10/31/14
;  

The variables I’m trying to produce are:

1. Signup month (easy to do)
2. A count of the total number of signups for that month (easy to do)
3. A overall count of additional products which included with sign up
4. A variable which has all add on product values (transposed from original dataset).
5. A count of the different products purchased on the startup date
6. A count of add on products purchased after the signup date that were purchased in the same month of the signup date 7.Then month variables which count the additional add on products by month

If I take just April, the output I'm looking for is something like this:

 data want ;
     length 
      Sign_up_Month $5
      Sign_up_count 8
      Initial_Products_total    8
      Products  $25
      Prod_Purchased_on_Signup  8
      AddPro_ April_After_SU 8
      May 8 June 8  July 8  August 8    September 8 October 8; 
     INPUT Sign_up_Month    $   
      Sign_up_count 
      Initial_Products_total    
      Products  $
      Prod_Purchased_on_Signup  
      AddPro_ April_After_SU    
      May   June    July    August  September   October; 
    datalines; 
    April   2   8   preferredhd_tv_estbd_dt     1                       
    April   2   8   ultimate_estbd_dt           2                           
    April   2   8   quant_estbd_dt                          1   
    April   2   8   FullyLoaded_estbd_dt    1                           1
    April   2   8   HB_estbd_dt            1                            
    April   2   8   Cin_estbd_dt    2                           
    April   2   8   time_estbd_dt   2                           
    April   2   8   router_estbd_dt     1       1               
    April   2   8   internet_estbd_dt                   1           
    ;

Below is the code I have for the first three vars in the output data set: signup_month, Sign_up_count, Initial_Products_total.

proc sort data=have; 
by ID signup_DT; run; 
proc transpose data=have out=have (drop=_LABEL_); 
by ID signup_DT; run; 
data have; 
set have; 
if signup_DT=COL1 then Initial_flag=1;run; 
proc sql; 
create table have as 
select  distinct
count( distinct ID) as Sign_up_count ,
month (signup_DT) as signup_month, 
sum (Initial_flag) as Initial_Products 
from have
group by month ( signup_DT) ; quit;

I'm having trouble creating the remaining vars: Prod_Purchased_on_Signup, AddPro_ April_After_SU and the counts by month.

I having been experimenting with arrays to try and accomplish this but I've been having trouble.

Answer 1

I'm not certain from your question what level of aggregation you want your counts to be at. But here is a solution if you are looking for summary for each distinct ID and sign-up date. This requires your original input sorted by ID signup_DT.

proc transpose 
    data = have 
    out = trans; 
    by ID signup_DT; 
run; 

/* Sort for by group processing and regular name order */
proc sort data = trans;
    by ID signup_DT _NAME_;
run;

data products (drop = _NAME_ COL1 i);
    set trans;
    /* For by group processing */
    by ID signup_DT;
    /* Get the signup month as a word */
    signup_month = put(signup_DT, monname.);
    /* Make the product list variable to prevent truncation */
    length Products $400.;
    /* Retain so we can add to the variables as we go down through the group */
    retain Products Sign_up_count signups_month0-signups_month4;
    /* Set up array reference for later month counts so we can loop */
    array som[5] signups_month0-signups_month4;
    /* Reset out new variables */
    if first.signup_DT then do;
        Products = "";
        Sign_up_count = 0;
        do i = 1 to 5;
            som[i] = 0;
        end;
    end;
    /* Add to the listt and count of sign up products */
    if signup_DT = COL1 then do;
        Sign_up_count + 1;
        Products = catx(" ", Products, _NAME_);
    end;
    /* Otherwise add to the later month counts by checking months seperating the dates */
    else do i = 1 to 5;
        if intck("month", signup_DT, COL1) = i - 1 then som[i] + 1;
    end;
    /* Only output once we have completed a group */
    if last.signup_DT and Sign_up_count  then output;
run;