2.2. Relationship between food waste and shelf life The connection between reduced food waste and extended shelf life was established by the use of multiple linear regression (MLR)to produce an equation describing how the relative waste (RW)depends on turnover (T), shelf life (SL) and minimum order size(MOS) for products where data on all these parameters are avail-able. In the MLR, the analysis was limited to including only food items with a shelf life shorter than 85 days. The calculations were based on 984 products, consisting of 92 cheese products, 258 dairy products 333 deli products and 311 meat products, and Eq. (1) was created from the MLR results (adjusted R2-value = 0.666). Log(RW) = 0.351 − 0.909 × Log(T) − 0.888 × Log(SL) + 0.156 × Log(MOS) (1)
The reduction in relative waste (RW) resulting from increased shelf life was calculated using this equation and then applied to the recorded waste of each product. The material included 3678 products, but 260 of these were excluded since they were stored either frozen or at room temperature and 391 were excluded since they had no recorded waste in any of the six supermarkets during all three study years. Since the material only included 984 products with values for all parameters studied (turnover, shelf life, mini-mum order size, waste), missing values were approximated based on similar products. Approximations had to be made for 1233 values of minimum order size and 1800 values of shelf life, which corresponded to 40% and 60% of the values, respectively. ‘Similarproducts’ in this case meant similar customer pack size and pack type when estimating minimum order size, and similar product type and brand when estimating shelf life. Potential reduction in RW was set by calculating a theoretical RW for every product and every temperature investigated. The original T and MOS were used in all calculations, but SL increased relatively to the original SL based on the sources cited in Section 2.1. This gave a theoretical relative waste value for every storage temperature and product investigated. The relative differences between the theoretical RW at starting temperature and theoretical RW at the target storage temperature were calculated and applied to the actual relative waste in the supermarkets to simulate the waste reduction potential of all products investigated.