SAS comparing sales trends over a quarter-CodePudding

I am looking to compare monthly sales data in one month to the following 3 months. For reference, i want to compare the sales rates in January to Feb, March and April sales and see the % growth or decline. I then want to do this for May over June/july/aug and then sept over OCT/NOV/DEC.

Example data i have is below. Not all SKUs sell year round (We dont sell Halloween branded items in March etc) so there is not 12 months of data for all SKUs.

SKU |  Sales | Month_number | Month
561     2           1          Jan
561     2           2          Feb
561     6           3          MAR
561     6           4          APR
191     5           8          AUG
191     10          9          SEP
191     10          10         OCT
191     12          11         NOV
.. ETC

Right now im using a lag statement to calculate month over month changes but cant think of a creative solution for calculating Jan vs March for a large dataset.

My current solution is to do the following

Data want;
Set have;
if month = 'Jan' then JanSales = Sales;
if month = 'MAR' then MarSales = sales;
MarJan = Marchsales/Jansales;
format Marjan percent8.2;
run;

This would create a new column called MarJan that is = 300%. in the case above. Now i could easily do that for every month but is there a cleaner, more efficient way to do this?

Thanks!

CodePudding user response：

We can do this by using a counter. On the first month, we'll store the sales value of the month. We'll then compare each subsequent month after that. For example, let's call n our counter and compare_month_sales the sales from the month we want to compare.

sales   month   n    compare_month_sales
2       Jan     1    2
2       Feb     2    2
6       MAR     3    2
6       APR     4    2 
5       AUG     1    5
10      SEP     2    5
...

When it's between 2 and 4, we have months that we want to compare to. Let's put this logic into action using by-groups. It looks like your data is sorted in a particular order, so we'll use the notsorted option.

data want;
    set have;
    retain compare_month_sales; /* Do not reset this variable at each row */
    by sku month_number notsorted;

    n 1;

    /* Create a counter for the number of months that have passed.
       If we've incremented to the 5th month, reset the counter */
    if(n = 5 OR first.sku) then n = 1;

    /* On the first month, get the comparison sale value */
    if(n = 1) then compare_month_sales = sales;

    /* Between The 2nd, 3rd, and 4th months will compare to the first */
    if(1 < n < 5) then pct = sales/compare_month_sales;

    format pct percent8.;

    drop n compare_month_sales;
run;

Output:

sku sales   month_number    month   pct
561 2       1               Jan     .
561 2       2               Feb     100%
561 6       3               MAR     300%
561 6       4               APR     300%
191 5       8               AUG     .
191 10      9               SEP     200%
191 10      10              OCT     200%
191 12      11              NOV     240%