Discrete Fourier Transformation for Seasonal-Factor Pattern Classification and Assignment

This paper introduces a data mining method to investigate the relationship between seasonal factors (SFs) and land use characteristics for urban areas in Florida through discrete Fourier transformation (DFT). First, DFT is applied to discover seasonal variation patterns, and two typical patterns were identified. Second, linear regression is used to determine influential variables, and a weighted similarity method derived from the amplitude of each DFT wave is applied for the SF assignment. The results obtained by DFT demonstrate promising assignment accuracy with a mean absolute percentage error of 4.27% for all data and 3.96% for the low seasonal household percentage subclass.