Periodic Reporting for period 1 - HomicidesLACcities (Understanding the drivers and impacts of homicides in 4 major Latin American cities.)
Período documentado: 2020-09-01 hasta 2022-08-31
This project has worked to better understand the drivers and effects of homicides in several major Latin American cities with of focus on the differential impacts on vulnerable groups, especially women. In particular, it has tackled the main two questions:
- What are the key correlates and drivers of homicides in these major cities?
- How are the lives of people affected by homicides in these cities, especially the most vulnerable?
However, as more research emerged about these topics, it was clear that the focus could not only be on the analyses of criminality and inequality in Latin America; but that it should be expanded into new areas (Senegal and Lebanon, for instance) and themes. Therefore, the research has also focused on the applications and implications of non-traditional data such as telecom operator data and satellite imagery for human development especially issues of violence, inequalities, and inclusion in the so-called “Global South”.
In addition to the research, that has brought new metrics, models, algorithms, insights, and data visualizations, the project also entailed a capacity and community building component and a dissemination and advocacy component to enhance its impact.
In summary, this project is important for society because it has yield an invaluable body of insights on the drivers and effects of homicides in several large Latin American cities, with a particular focus on women, as well as means of diffusion and discussions for greater impact. By its unprecedented breath and scope, this project constitutes a major advance for crime reduction, urban planning, and social policies in Latin America with applicability to other regions.
- What are the key correlates and drivers of homicides in these major cities?
- How are the lives of people affected by homicides in these cities, especially the most vulnerable?
However, as more research emerged about these topics, it was clear that the focus could not only be on the analyses of criminality and inequality in Latin America; but that it should be expanded into new areas (Senegal and Lebanon, for instance) and themes. Therefore, the research has also focused on the applications and implications of non-traditional data such as telecom operator data and satellite imagery for human development especially issues of violence, inequalities, and inclusion in the so-called “Global South”.
In addition to the research, that has brought new metrics, models, algorithms, insights, and data visualizations, the project also entailed a capacity and community building component and a dissemination and advocacy component to enhance its impact.
In summary, this project is important for society because it has yield an invaluable body of insights on the drivers and effects of homicides in several large Latin American cities, with a particular focus on women, as well as means of diffusion and discussions for greater impact. By its unprecedented breath and scope, this project constitutes a major advance for crime reduction, urban planning, and social policies in Latin America with applicability to other regions.
I have used non-traditional data, such as mobility data derived from cell-phones, to inquire whether different socio-economic groups in Mexico interact physically in the city or live in “parallel worlds”. A similar approach was used in six big Colombian cities (Medellin, Bogotá, Barranquilla, Monteria, San Andres, and Valledupar). In this case, I used a mix of traditional survey data, official statistics as well as non-traditional data, precisely cell-phone metadata from which indicators were computed by the data controller Telefonica-Movistar through a specific agreement to try to understand crime dynamics in these cities.
This new methodology has brought several relevant empirical results, among them, the following:
- a study about Mexico City showed one the very first empirical evidence of urban socio-spatial fragmentation leveraging mobility data, meaning the fact that different social groups access and use urban spaces in very different ways, and have few opportunities for physical social interactions.
- a study in multiple cities about the correlation of mobility data and crime produced significant results and showed that crime issues need to be tackled independently in each city, as there are no common patterns
- another study in Medellín, Colombia, showed that there is a relationship between the type and strength of social capital and crime patterns and trends.
Critical methodological innovations where also achieved:
- a study in Senegal related to the Sustainable Goal 7 proposed a novel spatio-temporal multi-target Bayesian regression model that provides accurate intercensal microestimates for household electrification and clean cooking fuel access by combining multiple types of earth-observation data, census, and surveys.
- research in 6 Colombian cities about crime dynamics using mobility, communication, and census data produced the development of estimates of 2 types of social capital commonly referred to as bringing capital and bonding capital based on cell-phone activity.
- the use of data on IOS device usage in Lebanon explains 90% of the out-of-sample variance in poverty across Lebanon.
- a new approach for measuring migration(s) as a demographic survival function, also relying on novel data sources (telecom operator data and geotagged social media data).
Moreover, the research has paved the way for additional research projects currently underway or about to start including on youth and gang violence in Port-au-Prince, Haiti, with funding from UNICEF Haiti, and climate-induced migration in Senegal with funding from the Belmont Forum.
This new methodology has brought several relevant empirical results, among them, the following:
- a study about Mexico City showed one the very first empirical evidence of urban socio-spatial fragmentation leveraging mobility data, meaning the fact that different social groups access and use urban spaces in very different ways, and have few opportunities for physical social interactions.
- a study in multiple cities about the correlation of mobility data and crime produced significant results and showed that crime issues need to be tackled independently in each city, as there are no common patterns
- another study in Medellín, Colombia, showed that there is a relationship between the type and strength of social capital and crime patterns and trends.
Critical methodological innovations where also achieved:
- a study in Senegal related to the Sustainable Goal 7 proposed a novel spatio-temporal multi-target Bayesian regression model that provides accurate intercensal microestimates for household electrification and clean cooking fuel access by combining multiple types of earth-observation data, census, and surveys.
- research in 6 Colombian cities about crime dynamics using mobility, communication, and census data produced the development of estimates of 2 types of social capital commonly referred to as bringing capital and bonding capital based on cell-phone activity.
- the use of data on IOS device usage in Lebanon explains 90% of the out-of-sample variance in poverty across Lebanon.
- a new approach for measuring migration(s) as a demographic survival function, also relying on novel data sources (telecom operator data and geotagged social media data).
Moreover, the research has paved the way for additional research projects currently underway or about to start including on youth and gang violence in Port-au-Prince, Haiti, with funding from UNICEF Haiti, and climate-induced migration in Senegal with funding from the Belmont Forum.
The project experimented innovative analytical approaches that leveraged new data sources and data science methodologies. Besides their empirical results, it is primarily these features that constitute progress beyond state of the art. These innovations are critical in a context where, with fewer than 7 years left before the 2030 milestone, there are still; major gaps in measuring some key SDG and human development indicators, and when some of the key resources that may bridge these gaps—novel data sources, data science and AI approaches—are widely under-utilized.
The project provided one the very first empirical evidence of urban socio-spatial fragmentation leveraging mobility data, meaning the fact that different social groups access and use urban spaces in very different ways, and have few opportunities for physical social interactions. Moreover, One, it showed that the combined use of socio-economic conditions, mobility information and physical characteristics of the neighbourhood account for a significant share of crimes, and do so better than any combination of these factors independently and that the ways in which these sets of characteristics relate to crime patterns and trends vary from city to city, suggesting that crime prevention and mitigation ought to be highly context-specific and subject to further investigation.
Furthermore, we studied a particular case in Medellín (Colombia), with a novel way to measure different forms of social capital from cell-phone activity. Crutially, this stud suggests that these different forms of social capital do matter for crime and that past criminal activity is a strong predictor of current criminal activity. Therefore, one of the most relevant societal impacts is that, through targeted policy intervention, one might break these vicious circles where past crime fuels future crime. However, one of the challenges of this study is the evience that factors associated with crime tend to be highly city-specific. Hence, there is the need for granular understanding of local dynamics and conditions to design appropriate interventions and investments to curb crime.
Moreover, I have developed and attempted to disseminate are that of “Human AI”, "learning feedback loops" and "human or society-in-the-loop." These concepts have become quite influential in policy circles and considered AI as an instrument to optimize or predict a number of tasks and an inspiration to build learning human systems and societies.
In addition, I co-founded an academic network of Excellence on AI and Sustainable Development named NAIXUS (Network for Artificial Intelligence, Knowledge and Sustainable development), which aims to promote academic research collaborations and exchanges at the inteserction of AI and the SDGs.
The project provided one the very first empirical evidence of urban socio-spatial fragmentation leveraging mobility data, meaning the fact that different social groups access and use urban spaces in very different ways, and have few opportunities for physical social interactions. Moreover, One, it showed that the combined use of socio-economic conditions, mobility information and physical characteristics of the neighbourhood account for a significant share of crimes, and do so better than any combination of these factors independently and that the ways in which these sets of characteristics relate to crime patterns and trends vary from city to city, suggesting that crime prevention and mitigation ought to be highly context-specific and subject to further investigation.
Furthermore, we studied a particular case in Medellín (Colombia), with a novel way to measure different forms of social capital from cell-phone activity. Crutially, this stud suggests that these different forms of social capital do matter for crime and that past criminal activity is a strong predictor of current criminal activity. Therefore, one of the most relevant societal impacts is that, through targeted policy intervention, one might break these vicious circles where past crime fuels future crime. However, one of the challenges of this study is the evience that factors associated with crime tend to be highly city-specific. Hence, there is the need for granular understanding of local dynamics and conditions to design appropriate interventions and investments to curb crime.
Moreover, I have developed and attempted to disseminate are that of “Human AI”, "learning feedback loops" and "human or society-in-the-loop." These concepts have become quite influential in policy circles and considered AI as an instrument to optimize or predict a number of tasks and an inspiration to build learning human systems and societies.
In addition, I co-founded an academic network of Excellence on AI and Sustainable Development named NAIXUS (Network for Artificial Intelligence, Knowledge and Sustainable development), which aims to promote academic research collaborations and exchanges at the inteserction of AI and the SDGs.