Predicting Sense of Community and Participation by Applying Machine Learning to Open Government Data

Community capacity is used to monitor socioeconomic development. It is composed of a number of dimensions that can be measured to understand issues possibly arising in the implementation of a policy or of a project targeting a community. Measuring these dimensions is thus highly valuable for policymakers and local administrator, though expensive and time consuming. To address this issue, we evaluated their estimation through a machine learning technique—Random Forests— applied to secondary open government data and determined the most important variables for prediction. We focused on two dimensions: sense of community and participation. The variables included in the data sets used to train the predictive models complied with two criteria: nationwide availability and sufficiently fine-grained geographic breakdown, that is, neighborhood level. Our resultant models are more accurate than others based on traditional statistics found in the literature, showing the feasibility of the approach. The most determinant variables in our models were only partially in agreement with the most influential factors for sense of community and participation according to the social science literature consulted, providing a starting point for future investigation under a social science perspective. Moreover, due to the lack of geographic detail of the outcome measures available, further research is required to apply the predictive models to a neighborhood level.