Northern American population data recovery from 1500AD to 1950AD as well as prediction using WASD neuronet with 513-year data

The recovery and prediction of Northern American population data, which are closely related to the future development of Northern America and even the whole world, have become significant subjects and captured great attention among sociologists as well as scientists. However, most of the relevant researches are just based on fertility, mortality or other individual quantifiable factors by traditional statistical models and thus lack all-sidedness in their results. As we know, the historical population data are the comprehensive reflection of the population development under the influence of all factors. In this paper, based on the past 513-year population data, a feedforward neuronet equipped with the weights-And-structure-determination (WASD) algorithm is thus constructed for the recovery and prediction of Northern American population data. Besides, the neuronet is activated by a group of Chebyshev polynomials (CP) of the first kind, which is named the CP neuronet in this paper. Moreover, the optimal normalization factor is found and utilized to improve the neuronet's performance. Due to the marvelous learning and generalization abilities of the presented CP neuronet, we recover the missing population data from 1500AD to 1950AD as well as draw up the Northern American population prediction for the next few decades.