Two customers walk into our agency. They state their preferences, share some contact information, and leave. When we summarize their preferences, we are amused--the vague terms are totally identical. Both the customers stated the following criteria:
- It shouldn't be too noisy
- There should be at least a restaurant and a bar nearby
- There should be some markets nearby
- There should be a park nearby, preferably with a playground
Additionally, both of them had a single, more specific request.
- Customer 1: I would like to place some solar panels on the rooftop, and use them efficiently.
- Customer 2: I don't have a car, and prefer to walk rather than use public transport. However, I like to travel, therefore, the train station and the bus station should be at a walking distance.
Our first task in such a case is to interpret the criteria, and translate them into the language of GIS. We have to create some exact steps from these preferences. Of course, having such vague criteria always leaves space to some subjectivity; therefore, the following interpretation accords to my experience:
- It shouldn't be too noisy: The results should be more than 500 meters away from industrial areas, and more than 200 meters away from busy roads (like motorways and highways) in terms of linear distance.
- There should be some amenities, shops, and a park with a playground nearby: The results should be less than 500 meters away from at least one park with a playground, bar, restaurant, and from two markets in terms of linear distance.
- Efficient solar panels: The aspect of the results' area should face South in the northern hemisphere, and face North in the southern hemisphere.
- Stations at walking distance: The cost of walking to those stations from the results must not exceed 15 minutes. If this criterion cannot be met (that is, the train and bus stations are too far away), or there are no results (that is, the intersection of the valid costs is too small to contain any results), we will use the union of the valid costs.
Implementing airport noise into the model is also possible, although our current data is not sufficient for this. To estimate airport noise pollution, we would need at least the paths of taking off and landing airplanes, and buffer them.