As the amount of waste decreases due to population decline and incineration facilities age simultaneously, a guideline for future waste generation needs to be considered when restructuring the facilities.
In this study, we conducted flurrying in the latent variable space of PLS regression for 198 municipalities across 2 prefectures and 4 counties in the Kansai region (594 cases in total) at 3 time points. We extracted 5 lifestyle-related and 4 business-related regional types and constructed a model incorporating these into a prediction formula. While the initial prediction was 15, a method using a common prediction formula across all regions with intercepts for each regional type resulted in a prediction of 0.281, achieving approximately twice the explanatory power. A hybrid method was adopted for estimation, correcting for structural changes based on actual values. Forward validation yielded an average error rate of 6.45% and a rate of 91.9% within ±15% of the error.
