Small businesses need accurate forecasts of how many customers will show up every day. This helps them plan (using queueing theory discussed in Chapter 45, “The Waiting is the Hardest Part”) their daily staffing needs. In this chapter, we use 4.5 years (June 2001 through the end of 2005) of daily customer counts at a branch of a national restaurant chain to develop a simple model that can be used to forecast daily supper customer counts. We will predict the daily customer count using equation (1). Our work is in the file
For example, suppose we want to predict the number of customers on Super Bowl Sunday, February 3, 2002. As we will see, this is the least busy day of the year for the restaurant. For this day:
- Day # = 219 (February 3, 2002 is the 219th day in the data set).
- Day of Week = Sunday.
- Week of the Year = 32. The restaurant classified Thursday, June 28, 2001 as the first week of the year. They wanted to avoid partial weeks, so in 2002, Thursday June 27 was the first day of the year; in 2003, Thursday June 26 was the first day of the year; in 2004, Thursday July 1 was the first day of the year; and in 2005, Thursday June 30 was the first day of the year. This classification results in 2003 having 53 weeks. February 3, 2002 is in the 32nd week of 2001.
- Special Day Adjustment will equal the adjustment for Super ...