conflict history regression model
We are finally getting better at predicting organized conflict
Incidents of conflict and protest, along with many other structural variables, are fed into constituent models. Input variables would include things like population density, GDP growth, travel time to the nearest city, proportion of barren land, years since independence, and type of government. Several different models, each of which uses a different method, compute a probability of conflict. Constituent models could be a conflict history regression model, natural resources model, and an aggregate machine learning model. The results from the constituent models get combined to produce a final risk score.