PREFECT : (Development and Evaluation of a Prediction Model for the Probability of Famine Catastrophes Using Data Analytics Techniques)

Starvation disasters occur in most cases due to a combination of different indicators, such as bad environmental conditions, crop failures or man-made influences such as corruption or armed conflicts. When various factors collide over a given time, the resources of a country are attacked and famine can occur. Different indicators may arrive at the same time, different factors favour other factors and a sequence of events happens and finally a famine situation occurs. When a famine has arrived, it is usually too late. The biggest problem for aid organizations in fighting famines is the lack of time to establish sustainable help and care. The aim of the study is to develop a mathematical forecasting model that calculates the likelihood of a famine so that adequate disaster relief can be ensured as far as possible before an emergency situation occurs. This predictive model will be created using data analysis methods and trained on the basis of historical data on the respective countries. Subsequently, a prediction accuracy of the model is to be validated. Once the accuracy of the model allows a good prediction of the training data sets, real predictions of the probability of famine can be calculated.