Modelo de sistema de alerta temprana para desbordamientos de arroyos en Barranquilla basado en la comunidad

Abstract

The present work seeks to design and create a community-based early warning model as an alternative to as an alternative for mitigating the disaster caused by overflowing streams in Barranquilla (Colombia). in Barranquilla (Colombia). This model is based on the contributions in social networks, which are consulted by means of the api of each social network and filtered according to their according to their location, The information collected is cleaned and debugged, and then with debugging, and then with natural language processing techniques to tokenize the texts, seeking to operate vectorize the texts, seeking to operate mathematically to find the vector similarity between processed texts, generating from between processed texts, generating in this way a classification between texts associated with stream associated with stream overflow and texts not associated with overflow. Additionally, the texts classified as stream overflow are processed again in order to obtain a location or assign a default one, in order to georeference these data on a map that georeferencing this data on a map that allows to associate the risk zone and visualize it in a web application, monitoring and decreasing and visualize it in a web application, monitoring and reducing the possible damage to the population. generated to the population. In order to choose the best classifier, 3 classification algorithms were selected (random forest, randomly generated and randomly generated). classification algorithms (random forest, extra tree and k-neigbor) were selected, which showed the best and R2 in reference to the data processed in the regressions performed. regressions. Finally, the three aforementioned algorithms were trained, found that the k-neighbor algorithm obtained 88 failures out of a test set of 400 tweets, being the one with the least number of failures. tweets, being this the one that obtained the least number of failures and selected for the proposed system.