The rise in machine learning has made the subject interesting for new types of uses. This Master thesis implements and evaluates an LSTM-based algorithm on the conflict forecasting problem. Data is structured in country-month pairs, with information about conflict, economy, demography, democracy and unrest. The goal is to forecast the probability of at least one conflict event in a country based on a window of historic information. Results show that the model is not as good as a Random Forest. There are also indications of a lack of data with the network having difficulty performing consistently and with learning curves not flattening. Naive models perform surprisingly well. The conclusion is that the problem needs some restructuring in order to improve performance compared to naive approaches. To help this endeavourpossible paths for future work has been identified.

The aim of this thesis is to attempt to address gaps in the forced displacement-conflict escalation literature, as well as in the literature on conflict forecasting. By posing the question To what extent can data on forced displacement improve accuracy of conflict escalation forecasts?, the aim is to explore the possible bidirectional relationship between forced displacement and armed conflict, as well as how such a relationship may be beneficial for conflict forecasts. Through utilizing Random Forest classifiers and regressors, two hypotheses are tested: 1) Increasing numbers of displaced persons are associated with escalating violence, and 2) Incorporating forced displacement data into conflict forecasting models can improve the accuracy of conflict timing prediction. The obtained results offer support for both hypotheses, in turn providing two contributions to the field of peace and conflict studies. First, that the relationship between displacement and conflict escalation is not strictly causal. Second, data on displacement magnitudes can improve conflict escalation forecasts.