Predictive analysis model to define behavioral patterns of landslide for early warning based on machine learning (#1571)

Authors

Diaz Amaya, Edgar

Alvarado Jimenez, Carlos Adrian

Lo Coronado, Lyang Jazmin

Abstract

Activations of streams, known as Landslide, are natural events that cause considerable damage to property and infrastructure, causing losses of around 5 billion dollars, which negatively impacts the economic stability of the country and the people. In this work, a predictive analysis model based on machine learning is proposed to predict the occurrence of Landslide in Chosica, Peru. The model was trained with data from hydrological and meteorological sensors and was able to identify behavioral patterns of the landslide with an accuracy of 80%. This study demonstrates that the proposed model is a viable tool that can perform an acceptable prediction rate with low error control.