Table 1.
The variables used in the development of regression models assessing the drivers of consumer food waste (note that some variables listed below are multifaceted due to the various product types addressed).
Avoidable food waste was the dependent variable and the others were the explanatory variables.
Fig 1.
a) Variable importance (the loss of accuracy of classification) as determined by the “Boruta” algorithm for the full variable set. Variables retained for model selection (those with high or medium importance) are highlighted in green and yellow. Shadow feature minimum, mean and maximum are highlighted in blue; b) with “Discard behaviours” removed from the variable set; c) with “Local authority” removed from the variable set; d) with both “local authority” and “discard behaviours” removed from the variable set.
Table 2.
Five plausible models (ΔAIC <2.0) were selected from the original set of 16,384 models.
Models were ranked by AICc (“:” indicates interaction terms). The averaged coefficients of the models are shown in S1 Table.