Figure 12From: Extracting complements and substitutes from sales data: a network perspectiveDistributions of the relative number of shared recipes of all product pairs \(\{rr_{ij}\}\) (“all”), of complementary pairs \(\{rr_{ij}: W_{ij}^{(c)} > 0\}\) (“com”) and of substitute pairs \(\{rr_{ij}: W_{ij}^{(s)} > 0\}\) (“sub”), where the range is chosen for visualisation purpose, while complementary pairs also have positive probabilities at values greater than 0.2Back to article page