This study analyzed changes in household consumption during a recession using data from Oita City and cluster analysis, a method within machine learning. The results revealed that non-working households tended to exhibit significant changes in consumption structure, although overall differences between the two groups were found to be limited.