Prediction of India's demographic and economic variables using the neural network auto-regression model

Abstract

Forecasting demographic and economic variables is an essential component of research that will help society and the government plan for the best or worst in the future. The data on demographic variables are collected from the Census of India and SRS publications, and economic variables are gathered from the Economic Survey of India from 1971 to 2020. The goal of this research is to forecast demographic and economic factors using the NNAR approach. Because statistical approaches such as the least RMSE training and testing values are used in the process of identifying this method, this research is expected to contribute to the neural network method coupled with the statistical method. The results of this study should be able to predict accurate demographic and economic characteristics. A Neural Network Auto-regression (NNAR) model is used to predict demographic and economic variables for the next ten years, with the best forecasting model being the NNAR (4,4), (4,4), (4,4), (11,6), (10,6), (10,6), (10,6), (5,6), (10,6), (6,4) models. The study's findings show that, except for GDP, all of the selected variables fit the NNAR model well and a comparison shows that the rural population is best fitted when using Mean Absolute Percentage Error (MAPE) when compared to the entire set of demographic and economic variables. The Rural population is the best-fitting model of the three populations; under-five mortality is well-fitting among vital rates; and age dependency ratio is the best forecasting in economic variables using mean absolute percentage error (MAPE).