0

I'm trying to find out which promotion was effective for Retail promotion data.

Data has the following column names: item_nbr,Promo_Type,sales,units,promo_units,promo_sales,fineline_nbr,dept_subcatg_nbr,dept_category_nbr,dept_catg_grp_nbr,acctg_dept_nbr,year,week

The formula to calculate is : Overall Lift = Lift in Promoted Item Sales + Halo – Cannibalization – Pull-forward

How do I go about calculating Overall lift(Halo, Cannibalization, Pull forward, lift) and baseline in R?

ArK
  • 20,698
  • 67
  • 109
  • 136
user7166577
  • 1
  • 3
  • 2
    What did you try (in terms of lines of code)? Why did it fail? Voting to close because the question, as it stands, has no use for any Stackoverflow visitor. – lukeA Nov 16 '16 at 10:00
  • 1
    Migrate to http://datascience.stackexchange.com – smci Nov 18 '16 at 11:43
  • 1
    This is on-topic at [DataScience.SE](http://datascience.stackexchange.com/), please migrate there rather than closing. cc: @lukeA – smci Nov 18 '16 at 12:04
  • 1
    @smci Seems I cannot change to migrate, just cancel my close vote. – lukeA Nov 18 '16 at 18:27

1 Answers1

0

You have to make decisions before your lift calculation is made. Are you looking for profit lift, unit lift, total sales dollar lift? What increase are you looking to measure?

And choose a baseline...is it the immediately proceeding x number of days before the promotion where x is the number of promotional days? Is it the number of sales made in the same period last year (without promotion) is it a periodic average (with all promotions removed from the set?

Then you simply take the promotional metric (say $ value in sales) minus the non-promotional baseline/ the non-promotional baseline . Example: ($15000-$2100)/$2100 or (12 sales - 4 sales)/ 4 sales.

You need to know your data set well enough to sort the specific sales by category and create those two classes of data properly to calculate lift.

sconfluentus
  • 4,693
  • 1
  • 21
  • 40