AN APPLICATION OF TIME DEPENDENT FOURIER AMPLITUDE MODEL ON FORECASTING THE UNITED STATE POPULATION

Abstract

This study smeared the Time Dependent Fourier Amplitude Model Approach to forecast the Population of the United States of America from 1790 to 2020 on a 10-year interval using Number Crunches Statistical software (NCSS). Results obtained using this procedure was matched with the results obtained in the other models: Malthusian, Logistics, and Logistics (Least Squares) Model. These models were matched using the goodness of fit (the coefficient of determination (R²), the sum of square error (SSE)), the Akaike information criterion (AIC), Bayesian information criterion (BIC), Mean Absolute Deviation (MAD), Mean Error (ME), and Mean Sum of square Error (MSSE), Results displays that the Time Dependent Fourier Amplitude Model has the highest R² and has the lowest SSE, AIC, BIC, ME, MAD, and MSSE. The normal probability plot of residual also forms a lined pattern. The Time Dependent Fourier Amplitude Model gives a statistically significant development in the data sets as compared to the earlier models and also is a suitable model for forecasting the United States population.