Biblioteca Digital de VanguardiaPredicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia
The use of technology to prevent and respond to citizen security challenges is increasingly frequent. However, empirical evidence has been concentrated in major cities with large amounts of data and local authorities' strong capacities. Therefore, this investigation aims to capture a...
| Autores principales: | , , |
|---|---|
| Formato: | Revistas |
| Lenguaje: | Español |
| Publicado: |
FLACSO - Sede Ecuador
2022
|
| Acceso en línea: | https://revistas.flacsoandes.edu.ec/urvio/article/view/5395 |
| _version_ | 1809907060450525184 |
|---|---|
| author | Gelvez Ferreira, Juan David Nieto-Rodríguez, María-Paula Rocha-Ruiz, Carlos-Andrés |
| author_facet | Gelvez Ferreira, Juan David Nieto-Rodríguez, María-Paula Rocha-Ruiz, Carlos-Andrés |
| author_sort | Gelvez Ferreira, Juan David |
| collection | Revista |
| description |
The use of technology to prevent and respond to citizen security challenges is increasingly frequent. However, empirical evidence has been concentrated in major cities with large amounts of data and local authorities' strong capacities. Therefore, this investigation aims to capture a series of policy recommendations based on a machine learning crime prediction model in an intermediate city in Colombia, Bucaramanga (department of Santander). The model used signal processing for graphs and an adaptation of the TF-IDF text vectorization model to the space-time case, for each of the cities’ neighborhoods. The results show that the best crime prediction outcomes were obtained when using the models with spatial relationships of graphs by weeks. Evidence of the difficulty in predictions based on the periodicity of the model is found. The best possible prediction (with available data) is weekly prediction. In addition, the best model found was a KNN classification model, reaching 59 % of recall and more than 60 % of accuracy. We concluded that crime prediction models are a helpful tool for constructing prevention strategies in major cities; however, there are limitations to its application in intermediate cities and rural areas in Colombia, which have little statistical information and few technical capabilities. |
| format | Revistas |
| id | urvio-article-5395 |
| institution | URVIO. Revista Latinoamericana de Estudios de Seguridad |
| language | Español |
| publishDate | 2022 |
| publisher | FLACSO - Sede Ecuador |
| record_format | ojs |
| spelling | urvio-article-53952022-10-07T17:04:27Z Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia Prediciendo el crimen en ciudades intermedias: un modelo de “machine learning” en Bucaramanga, Colombia Gelvez Ferreira, Juan David Nieto-Rodríguez, María-Paula Rocha-Ruiz, Carlos-Andrés análisis de datos Colombia crimen prevención del crimen Policía crime crime prevention Colombia data analysis police The use of technology to prevent and respond to citizen security challenges is increasingly frequent. However, empirical evidence has been concentrated in major cities with large amounts of data and local authorities' strong capacities. Therefore, this investigation aims to capture a series of policy recommendations based on a machine learning crime prediction model in an intermediate city in Colombia, Bucaramanga (department of Santander). The model used signal processing for graphs and an adaptation of the TF-IDF text vectorization model to the space-time case, for each of the cities’ neighborhoods. The results show that the best crime prediction outcomes were obtained when using the models with spatial relationships of graphs by weeks. Evidence of the difficulty in predictions based on the periodicity of the model is found. The best possible prediction (with available data) is weekly prediction. In addition, the best model found was a KNN classification model, reaching 59 % of recall and more than 60 % of accuracy. We concluded that crime prediction models are a helpful tool for constructing prevention strategies in major cities; however, there are limitations to its application in intermediate cities and rural areas in Colombia, which have little statistical information and few technical capabilities. El uso de tecnología para prevenir el crimen es una práctica cada vez más frecuente. Sin embargo, la evidencia se ha concentrado en ciudades principales, que cuentan con gran cantidad de datos y mejores capacidades locales. El objetivo de esta investigación es presentar los resultados de un modelo de “machine learning” para predecir el delito en Bucaramanga, una ciudad intermedia de Colombia. Se utilizó el procesamiento de señales para grafos y una adaptación al caso del modelo de vectorización de texto TF-IDF. Se identificó que los mejores resultados en la predicción del crimen se dieron con modelos espaciales de grafos por semanas. Además, encontramos evidencia de que existen diversas dificultades de predicción, en dependencia de la periodicidad del modelo. La mejor opción posible (con los datos disponibles) es una periodicidad semanal. El mejor modelo encontrado es un KNN de clasificación, que alcanza un 59 % de exhaustividad(recall) y más de 60 % de exactitud (accuracy.). Concluimosque los modelos de predicción del delito constituyen una herramienta útil para construir estrategias de prevención en ciudades principales; sin embargo, existen limitaciones para su aplicación en ciudades intermedias, que cuentan con poca información. Abstract The use of technology to prevent and respond to citizen security challenges is increasingly frequent. However, empirical evidence has been concentrated in major cities with large amounts of data and local authorities' strong capacities. Therefore, this investigation aims to capture a series of policy recommendations based on a machine learning crime prediction model in an intermediate city in Colombia, Bucaramanga (department of Santander). The model used signal processing for graphs and an adaptation of the TF-IDF text vectorization model to the space-time case, for each of the cities’ neighborhoods. The results show that the best crime prediction outcomes were obtained when using the models with spatial relationships of graphs by weeks. Evidence of the difficulty in predictions based on the periodicity of the model is found. The best possible prediction (with available data) is weekly prediction. In addition, the best model found was a KNN classification model, reaching 59 % of recall and more than 60 % of accuracy. We concluded that crime prediction models are a helpful tool for constructing prevention strategies in major cities; however, there are limitations to its application in intermediate cities and rural areas in Colombia, which have little statistical information and few technical capabilities. Pesquisas acadêmicos mostram que modelos preditivos possuem ótimos resultados na prevenção de crimes. O uso de tecnologia para prevenir e responder aos desafios da segurança dos cidadãos é cada vez mais frequente. Entretanto, a evidência tem sido concentrada nas grandes cidades com elevadas quantidades de dados e forte capacidade das autoridades locais. Em vista disso, este documento visa captar uma série de recomendações baseadas em um modelo preditivo de crime em Bucaramanga, cidade colombiana de tamanho intermediário localizada no departamento de Santander. Este modelo utilizou, para cada um dos bairros da cidade, um processamento de sinal de gráficos e uma adaptação do modelo de vetorização de texto TF-IDF para o caso espaço-tempo. Nossos resultados mostram que os preditores de crime que obtiveram os melhores desempenhos foram atingidos quando utilizados modelos com relações espaciais de gráficos por semana. Da mesma forma, o documento conclui que os modelos de previsão de crime são uma ferramenta útil para a construção de estratégias de prevenção nas grandes cidades; no entanto, há limitações para sua aplicação em cidades de porte médio e áreas rurais na Colômbia, que dispõem de escassas capacidades técnicas e tampouco de informações estatísticas. Ademais, é necessário analisar a fundo as implicações éticas do uso de ferramentas que operam por inteligência artificial. FLACSO - Sede Ecuador 2022-09-30 info:eu-repo/semantics/article info:eu-repo/semantics/publishedVersion text/html text/html application/pdf application/zip https://revistas.flacsoandes.edu.ec/urvio/article/view/5395 10.17141/urvio.34.2022.5395 URVIO. Revista Latinoamericana de Estudios de Seguridad; No. 34 (2022): Urvio. Revista Latinoamericana de Estudios de Seguridad (septiembre-diciembre); 83-98 URVIO. Revista Latinoamericana de Estudios de Seguridad; Núm. 34 (2022): Urvio. Revista Latinoamericana de Estudios de Seguridad (septiembre-diciembre); 83-98 1390-4299 1390-3691 10.17141/urvio.34.2022 spa https://revistas.flacsoandes.edu.ec/urvio/article/view/5395/4193 https://revistas.flacsoandes.edu.ec/urvio/article/view/5395/4194 https://revistas.flacsoandes.edu.ec/urvio/article/view/5395/4204 https://revistas.flacsoandes.edu.ec/urvio/article/view/5395/4261 Derechos de autor 2022 Juan David Gelvez Ferreira, María-Paula Nieto-Rodríguez, Carlos-Andrés Rocha-Ruiz |
| spellingShingle | Gelvez Ferreira, Juan David Nieto-Rodríguez, María-Paula Rocha-Ruiz, Carlos-Andrés Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title | Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title_full | Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title_fullStr | Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title_full_unstemmed | Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title_short | Predicting Crime in Middle-Size Cities. A Machine Learning Model in Bucaramanga, Colombia |
| title_sort | predicting crime in middle-size cities. a machine learning model in bucaramanga, colombia |
| url | https://revistas.flacsoandes.edu.ec/urvio/article/view/5395 |