Optimizing Expected Profit Under Flexible Allocation of Campaign Days and Budget for Management of Shopping Centers: A Machine Learning Approach

In retail businesses, a sales campaign is typically organized over segments of consecutive days within a certain period so as to maximize total sales in that period. It is also common to design a sales campaign so that good-sales-days of the previous year would be designated as sales campaign days, with the expectation that the campaign effect could further enhance the potential of good-sales-days. Sharkasi, Sumita, and Yoshii (2015) showed that, contrary to these common practices, expected total sales can be increased by reallocating sales campaign days in a more flexible manner. Their paper, however, focused exclusively on expected total sales without regard to expected profit and did not consider the sales campaign budget. The purpose of the current paper is to fill this gap by incorporating the sales campaign budget per day, where the expected total sales would be increased as a concave function of the budget increment. Furthermore, expected profit rather than total expected sales is optimized. Based on a real dataset from a shopping center in Tokyo, this study shows that expected profit could be increased considerably with no additional cost using the optimization approach.