Context
We will briefly introduce the three trending concepts that justify this thesis: Smart Cities, Social Impact Bonds and Data for Good Initiatives. They do not stand alone, but are are closely related. Together, they are the basis of a public policy data-driven ecosystem in which this thesis is built upon. The first concept, Smart Cities, does not have an unanimous definition [cite], but it can be summarised as an urban area that uses data generated by itself [city-data] to manage assets and resources efficiently [cite]. This concept aggregates several types of projects of city governance[cite]. Each of them builds different types of governance models using specific kinds of city-data to different goals. But the overlaying objective is always the same: rethink the government structure using a data-based paradigm. The second concept, Social Impact Bonds [SIB], 'are public-private partnerships that drive resources toward effective social programs that measurably improve lives' [cite], which was created to overcome previous contrcts that did not have the social improvement as a main goal. The SIBs is a kind of bond that pays more for their investors if a social indicator is improved in a given time frame.Currently, there are SIBs on place [cite] , but they usually lack of proof whether the company was responsible for indicator change [cite]. Finally, Data for Good Initiative, is tech community trend of using public or private data for the 'good' of society [cite]. This trend gained momentum that led private companies to open their users database, with private concerns, to positively impact society. Currently, Waze and Facebook are experimenting this model with the Center of Technology and Society - FGV, just to mention an example.
In this context, the Smart Cities are the underlying paradigm that changes how the city makes decisions. It shifts the purely political and heuristic city decisions to a more technical and optimal approach. One of the tools available to put this paradigm in practice is a SIB, since it is concerned in proofing the effectiveness of a contract. However, usually public generated data has disadvantages, it can be either hard to get automatically or unstructured; also, it can exist in a city, but not in another. One way to solve most of those problems is to use company based data, that are usually well structured and has broad geographic range. Nowadays, this is possible due to the Data for Good trend that moved companies to open their databases.
Also, road infrastructure is essential to guarantee safe mobility which is a drives of the PIB of a city. Not to mention the loss of life that is caused by these issues.
Therefore, we propose a methodology that allow SIB to evaluate the effectiveness of a public-private contract.. The case pretended to be studied is how to measure citizen satisfaction of road infrastructure maintenance using Waze data. This dataset was chosen due its geographic permeability (coverage?), covering all major cities around the Globe, and also its "engagement data" collected through interaction with users. In this work we adopt an open source licence so that the methodology can be freely used, criticized and extended by interested parties around the World.
Goals
The main goal of this thesis is to propose a method to measure the level of complaints about road infrastructure in a region of a city using geospatioal and temporal data (Waze). Also, suggest a experiment to test whether a social bond contract [SIB] decreases the level of complaints in a region of a city. This unfolds on several subgoals to ensure the reliability and scalability of the method. The subgoals are:
1. Study bias present in the Waze data for the proposed methodology, highlighting its limitations.
2. Propose a consistent way to divide the city in regions.
3. Propose a consistent way to aggregate complaints geographically.
4. Propose a method to measure the variation of complaints, that has to be controlled by global user variation on the full city and nearby regions.
5. Propose a experiment to implement the social bond contract in a city (or several) that can be falsified.
Underlying the mathematical methods, the procedural goal is to develop an modular solution that provides errors for each part.