yeg13/Malawi

Scripts from this repo were used to explore disaster-prediction metrics. Abstract from this paper, An approach to improving early warning systems: Using spatially and temporally rich data to predict food insecurity crises in Malawi, is as follows: "While the causes of famine are complex and include institutional failure and conflict, policymakers and practitioners often lack the necessary information to identify food insecure populations and, therefore, to effectively allocate scarce resources to mitigate hunger. As a result, humanitarian responses tend to trail the onset of food security crises. Our paper aims to enhances the early warning for food insecurity crises. The last decade has seen a dramatic increase in the available quantity and quality of data related to food security, rainfall, and prices. The full potential of these data has not yet been exploited, and they are often evaluated in isolation. We connect disparate datasets and use variation over space and time to build an evidence-based approach to estimating food security outcomes. We utilize price and precipitation data collected in near realtime to predict sub-national food insecurity, as assessed by the Integrated Food Security Phase Classification (IPC). By overlaying mapped measures by season and analysing changes over time, our model has the potential to predict future sub-national food insecurity at relevant spatial and temporal scales. We find that contemporaneous price and measures of the prior year’s precipitation consistently predict the IPC in Malawi. We then compare the IPC food security evaluations to household measures of food insecurity in Malawi to ask whether our simple price and weather model improves the prediction of food security outcomes. We find that the IPC is strongly associated with the reduced Coping Strategies Index, a household measure of food insecurity, indicating that the IPC method does successfully capture food security. This model is a crucial step toward improving …