Pinheiro, Flávio Luís PortasTorres-Sospedra, JoaquínPainho, Marco Octávio TrindadeGuzzardo, Frank2022-03-162022-03-162022-03-02http://hdl.handle.net/10362/134614Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial TechnologiesIn the last few decades there has been a steady increase in intrastate conflict around the globe. In response, there is a rising need for actionable information for national and international stakeholders to better forecast and mitigate the effects of intrastate conflict. The Sahel region is especially vulnerable to intrastate conflict suffering a multidimensional crisis that includes climate change, food insecurity, and the proliferation of armed conflict. This study seeks to explore the feasibility of producing a heuristic machine learning model utilizing open-source data to predict localized intrastate conflict events on a regional scale using the random forest regression algorithm. The model includes data from 2007 to 2020 selected from multiple sources to create 17 features representing real-world phenomena to predict conflict occurrence. A unified spatial data structure consisting of quadratic grid cells was used for local-level analysis. Implementing a 10-fold cross-validation method, the model performed well with an RMSE of 1.394 and an R2 of .95. There was an improvement of 76% from the baseline model.engSahelConflictRandom ForestPredictionmaster thesis202965929