My work for this project can be found in this notebook. It includes responses to the exercise's questions, my own analysis and that analysis' accompanying visualizations.
When considering restaurant coupons, which drivers are most likely to accept? And does this acceptance rate vary from inexpensive restaurant coupons to expensive restaurant coupons? I examine three filters that I surmise will show the highest acceptance rates.
Inexpensive restaurant coupons are accepted at a higher rate than expensive restaurant coupons. In both cases, drivers with specific restaurant-visiting habits accept at higher rates.
Ultimately, while restaurant-visiting habits turned out to affect chances of coupon acceptance for restaurant coupons, the difference it represented was modest and samples with the highest acceptance rates also proved fairly small. The data illustrates the risk of overemphasizing the significance of appealing subsets.
A business use case for this analysis would be to target restaurant coupon offers to people whose habits align with the highest acceptance restaurant visitation habits cited. Coupons for inexpensive restaurants should be offered to people who visit such businesses more than 8 times a month, while coupons for expensive restaurants should be offered to people who visit such businesses about 1 to 3 times a month.
For further analysis, more combinations of filters could be experimented with. Some of the combinations I tried yielded tiny sample sizes, sometimes with 100% or 0% acceptance rates, but maybe there's a combination the sample of which is big enough to justify inference but precise enough to describe a specific, ideal demographic.