Fitting Machine Learning Models for the Identification of Social Vulnerability in the Event of Political Instability in Nigeria

Abstract

Due to the high rate of poverty and the unequal distribution, social vulnerability is extremely common in rising economies around the world, including Nigeria. As a result of political instability in Nigeria, this research study's machine learning has been suitably fitted to identify potential social vulnerability. The outcomes of the machine learning optimizations indicate that a high incidence of social inequality, political unrest, natural disasters and agricultural instability will probably all contribute to the high degree of social vulnerability in Nigeria. The results of the predictor variables' contribution to the likelihood of high social vulnerability in Nigerian communities indicate that, at 100% and 74.8%, natural disasters related to flooding and political grievances respectively account for the majority of Nigeria's high level of vulnerability. Surpassing the logistic regression method, support vector machine, and random forest, the artificial neural network (ANN) attained the maximum prediction accuracy of 85% with a precision of 82%, according to the model performance evaluation. Therefore the best model for forecasting high social vulnerability in Nigerian currently, is the ANN. In order to reduce the high level of social vulnerability, the Nigerian government should establish an all-inclusive government that will resolve political grievances among citizens and also establish an efficient security network that will combat the country's current high level of insecurity. In the event that political instability, the government should then embrace the use of machine learning models for the future prediction of social vulnerability.