Forecasting automotive waste generation using short data sets: case study of Lithuania

Abstract

There were 1.83 million cars and average passenger car age was 18 years in Lithuania in 2013. Increasing number of cars has an insignificant effect on car age change but it is contrary to automotive waste, both hazardous and non-hazardous, that accumulates during vehicle exploitation and after it ends. The aim of this study was to assess different mathematical modelling methods abilities to forecast non-hazardous and hazardous automotive waste generation. Artificial neural networks, multiple linear regression, partial least squares, support vector machines, nonparametric regression and time series methods were used in this research. Results revealed that nearly perfect theoretical results in both cases can be reached by smoothing splines and other nonparametric regression methods. It is very doubtful that results would be so precise using data outside of currently used data set range and due to this reason further testing using 2014–2015 data is needed.