A prediction model of PET bottle and can collections by a recycling bin at public space

Abstract

This study surveyed collection performances of PET bottle and can using smart recycling bins (SmaGO® FORCETEC Co.) at Omotesando, Tokyo in July, 2022. Coupled with pedestrian traffic flow data (KDDI Location analyzer®), a model to predict waste collection of single recycling bin was developed and validated. Because waste collection data followed logarithmic normal distribution, a log-normal distribution-based model as a function of the distance to a recycling bin was tested. This model could not predict waste collection performances sufficiently. Because waste collections were significantly influenced by weather, temperature, and day, the independent variable was modified to include them as well as logarithmic transform of the distance. The modified model predicted waste collection performances better. This study also found that the effects of sunny weather and low temperature should be evaluated more appropriately to improve model underestimation. On the other hand, different factors rather than weather, temperature, and day might have caused model overestimation.