“We try to solve very complicated problems without letting people know how complicated the problem was. That's the appropriate thing.”Jonathan Ive
Parkroulette is based on the public dataset of parking tickets issued in San Francisco. For each ticket, the time, location, violation type, amount, and officer badge number are known. In total, about 3M tickets spanning 2.5 years are included in Parkroulette's database. That's about 4 tickets for every resident of San Francisco.
The parking tickets alone are not enough to tell what the chances are of getting a ticket if you park somewhere. You need to know the chance of a patrol passing by. Parkroulette reconstructs historical patrols by connecting the dots of tickets issued by each SFTMA badge number.
Altogether, the patrols form a dense criss-cross of lines that cover the entire map of San Francisco. When you ask Parkroulette for the chance of a ticket at a particular address, Parkroulette queries a GIS database to get the sum of patrols that passed within the given radius of the address, grouped by day and hour.
Parkroulette makes a second query to get the ticket information, so you can see the reasons why people get tickets in an area. The "costs" heatmap display merges the patrols and tickets information. The cost shown in each cell is the average ticket cost multiplied by the chance of a patrol.
Read about the known issues that affect Parkroulette.