Contingency table with 2 different time series I am trying to put the two time series of my Excel spreadsheet (see link below), into a contingency table.
https://www.dropbox.com/s/2f96oylxj97fuih/example.xls
The first series is the number of subscriptions per hour for 24 hours. The second series consists of the hours when ads for different radios stations have been displayed (e.g. ad for Channel 1 at 2:00PM).
I have doubts about the feasibility of putting these 2 time series into a contingency table. Is this reasonable?
 A: At this late date it's not likely that answers will be obtained for any of these observations and queries. 
A media analyst would instantly recognize the "dependent" vs "independent" nature of the information, i.e., subscriptions are dependent on radio advertising. That analyst would also know that advertising is typically sold by the daypart, defined roughly as fringe (12 am to 6 am), early morning (6 am to 10 am), daytime (10 am to 3 pm), and so on. So, collapsing hourly information to the daypart would be useful.
Key pieces of information are missing from the description. E.g., no time frame is given for this data. Is it for a single day? Is it aggregated over multiple days? Is it a time series? What's the duration or span of time? Given answers to these very basic concerns, considerations such as ad weight, flighting, pricing, seasonality as well as the editorial environment of the radio stations' audience could potentially be evaluated for their impact on subscriptions.
Another consideration is the unit of analysis for the advertising. Is it dollars spent? CPM (cost per thousand listeners reached)? Etc.
Also highly relevant is the radio station where the ads were aired. These are likely to be distributed by region (maybe, maybe not) but based on the description, it sounds like the subscription information is aggregated and not available at a regional level. This would mandate also aggregating the advtg information up from the station level. 
These are a few considerations that would complicate putting this information into a single "contingency table." However, multiple tables are possible breaking out the data temporally, regionally, by station, etc. 
In addition, appropriate predictive models could be built as a function of the framing of the information which would predict subscriptions by daypart and any other relevant factor. The output from these models could then be represented, maybe not as a contingency table, but certainly as a pivot table comparing predicted with actual subscriptions as a function of these underlying factors.
A: After much thinking and much asking around, the only sensible answer is that we cannot have a contingency table as we have TWO time series.
I hope it helps all the learner in stats.
