Classification of Consumption Level in Developing Countries for Time Series Prediction Using a Hierarchical Nested Artificial Neural Network Method

A significant factor in the negative impact of the operation of a developing country is the change in global economic conditions, combined with the possibility of the materialization of one or many of the risks that threaten its economic environment. Through the analysis of time series, it is possible to identify patterns or similarity in the information over time, as well as to predict future values, for which it is necessary to have technological tools that contribute to the timely integration of the results of the study carried out. Thus, our proposal consists of a hierarchical method of nested neural networks to classify levels of consumption in developing countries for time series prediction. The main contribution consists in using multiple unsupervised neural networks to classify each of the countries and subsequently supervised neural networks to predict their future values. The results of the simulation carried out show the advantages of using the proposed method, with which it is possible to make individual comparisons or by segments of countries with similar historical data.