The time series data of infections of coronavirus disease 2019 (COVID-19) typically contain various important components. Therefore, rigorous analysis of these components is essential to prevent and control the spread of new infectious diseases. This study investigates the diffusion routes of COVID-19 in Japan’s Kanto region by analyzing daily time series data of new COVID-19 cases for each prefecture. Utilizing a decomposition procedure based on the extended moving linear model approach, we isolated the trend, weekly variation, and cyclical components within the data. Thus, this study contributes to the development of statistical methodology in this field through presentation of a new approach for decomposing components of daily time series data of new COVID-19 cases in the Kanto region of Japan. Moreover, we conducted a correlation analysis between the data for Tokyo and its adjacent prefectures, incorporating time lags derived from trend components. The analysis indicated that the decomposed trend component captured smooth, long-term variations, while the cyclical component exhibited short-term periodic fluctuations. The results suggest potential transmission routes of COVID-19 from Tokyo to its neighboring prefectures.