Interpolation Method for Synthesized Household Compositions Between Census Years

In this paper, we propse an interpolation method for synthesized household compositions between census years by optimizing transitional populations using a genetic algorithm (GA). synthesized household is a set of artificial individual attributes of each household including household members that are synthesized based on census data. Since the census is conducted every five years in Japan, the current synthetic population for Japan does not include data for years when the census is not conducted. For these gap years between census years, there is a need to generate interpolated data to reflect the demographic trends during that period. In our study, we interpolate data at five-year intervals based on population transition data for years. We enumerate issues related to the continuity of the current synthetic population for each year in Japan and propose solutions to these issues. Population migration data between cities and prefectures are published annually by the Japanese Statistics Bureau. To make the data interpolation more accurate, we employ GA to optimize the population by gender and five-year age groups such as “Age 0 to 4”, “Age 5 to 9” and so on. The interpolated data generated by this study reflect more accurately actual population dynamics and will be useful for future population projections and policy making. In addition, simulations that were previously possible only at five-year intervals will now be possible at one-year intervals.