Application of Evolving Self-Organizing Maps for Analysis of Human Adverse Events in the Context of Complex Socioeconomic Infrastructure Interactions

The main objective of this study was to investigate complex human socioeconomic infrastructure interactions and information on past human adverse events (AE) in an active war theater in order to predict future AE in a given geographical region. Human AE were defined as those security-related events that threatened human lives. Human socioeconomic infrastructure development data were derived by integrating three different datasets from different sources based on the United States Agency for International Development database. Using empirical data obtained from the country of Afghanistan from 2002 to 2010, we applied evolving self-organizing maps (ESOM) to forecast future patterns of such AE. Records from 2003-2009 were used as training data, while records from year 2010 were used to test the efficacy of ESOM in predicting AE. The socioeconomic data, dates, and geographical location information was used as input for the trained model. ESOM algorithm with supervised learning was effective in understanding future patterns of AE in a war region. The results also showed the possibility of predicting future AE based on the incomplete information pertaining to the geographical location, recent history of AE in the specific region of the country, and relevant socioeconomic infrastructure development data. The differences in applying the classical self-organizing maps and ESOM approaches for modeling of complex human socioeconomic infrastructure interactions were also discussed.