Waste container weighing data processing to create reliable information of household waste generation

Household mixed waste container weighing data was processed by knowledge discovery and data mining techniques to create reliable information of household waste generation. The final data set included 27,865 weight measurements covering the whole year 2013 and it was selected from a database of Helsinki Region Environmental Services Authority, Finland. The data set contains mixed household waste arising in 6m(3) containers and it was processed identifying missing values and inconsistently low and high values as errors. The share of missing values and errors in the data set was 0.6%. This provides evidence that the waste weighing data gives reliable information of mixed waste generation at collection point level. Characteristic of mixed household waste arising at the waste collection point level is a wide variation between pickups. The seasonal variation pattern as a result of collective similarities in behaviour of households was clearly detected by smoothed medians of waste weight time series. The evaluation of the collection time series against the defined distribution range of pickup weights on the waste collection point level shows that 65% of the pickups were from collection points with optimally dimensioned container capacity and the collection points with over- and under-dimensioned container capacities were noted in 9.5% and 3.4% of all pickups, respectively. Occasional extra waste in containers occurred in 21.2% of the pickups indicating the irregular behaviour of individual households. The results of this analysis show that processing waste weighing data using knowledge discovery and data mining techniques provides trustworthy information of household waste generation and its variations.