Towards reliable recurrent disaster forecasting methods: Peruvian earthquake case

We are interested in recurrent disaster forecasts; these are events such as annual cyclones in the Caribbean, earthquakes along the Ring of Fire and so on. These crises, even small- or medium-sized, are, in fact, critical for the emergency response of humanitarian organizations inasmuch as the sum of casualties and losses attained are as deadly as those that are considered exceptional. The aim of our research is to show that it is possible to use traditional forecasting methods such as: causal methods (which include the use of linear regression functions, non-linear, multivariate, etc.), time series (which include simple moving average, weighted moving average, exponential smoothing, trend-adjusted exponential smoothing, etc.) and so on, if the historical data keeps, among other criteria, its patterns, frequency, and magnitude, in a sustainable manner. Finally, an example to forecast recurrent earthquakes in Peru is presented.