Analysis of the Level of Geographic Criminal Risk Oriented to Women

In this research, a methodology is presented to estimate and visualize the level of insecurity in geographical areas, using spatio-temporal analysis and data mining techniques focused in crimes against women. Data sources used are from official crime reports and news media, classified as: femicide, armed robbery and rape. Data were extracted using web scrapping in digital media publications and were collected from open databases provided by the Mexican government. It is distinguished the crimes reports against women, based on the fact they are classified in news media as feminicides while in open data appears as gender violence. The case study is focused on the municipality of Ecatepec de Morelos, State of Mexico due to its high crime density according to official reports. The results show a geographical and temporal description of the crime behavior in space and time. It allows estimating the level of risk for women in a geographical area at the suburb level and a day granularity. The approach was tested using a web tool that facilitates decision-making based on the representativeness of crime behavior, the dataset includes around 20,000 records in the years from 2018 to 2019. Future work includes the integration of the machine learning process to possible forecast and discover correlations, validations and filtering of possible fake news is other possible direction, and include other municipalities of Mexico.