I am going to analyze some data for an intermittent operation using R. Let's say I operate a Xmas tree stand from Black Friday to Christmas Eve every year. Let's say I operate 150 different Xmas tree stands every year. Let's say I have data for revenue-per-day for these 150 Xmas tree stands for the last 6 years. I want to analyze this data to figure out how to maximize my profit.
Trends that I expect to exist include
- Daily trends in volume (higher volume on weekends, etc)
- Trend of # of days before Xmas (start low, max somewhere then decrease probably)
- Which day of week Xmas is probably has an affect
- Year over year (should show Great Depression of 2008, then recovery etc)
I want to know how many stands to operate, and if I would do better with half the stands but move them around to different locations on different days to capture max revenue with minimum costs. For instance, I might find that for year when Xmas is on Wednesday, Location A should not be manned until 3 weeks before Xmas, should be manned on Thursday, Friday and Saturday for that week, and only on Tuesday and Saturday for the next 2 weeks. Bad example, but hopefully shows what I'm looking for.
I want to try this in R. My limited understanding of time series is stumped when I have ~335 days between data sets which require a resolution of 1 day. Should I instead add dummy variables of day of week, days before Xmas, year and just try to fit on them?
Any insight would be appreciated.