Table 2 The correlations between the scores from sales data, \((W^{(c)}_{ij})\), \((W^{(s)}_{ij})\), and the measures from the flavour compound and recipe data, \((rf_{ij})\), \((rr_{ij})\) (where the superscripts \((c)\) and \((s)\) below denote the values restricted to \(\{(i,j): W^{(c)}_{ij} > 0\}\) and \(\{(i,j): W^{(s)}_{ij} > 0\}\), respectively), \((W^{(cr)}_{ij})\), \((W^{(sr)}_{ij})\)
From: Extracting complements and substitutes from sales data: a network perspective
Score pair | Pearson | (p-value) | Spearman | (p-value) |
|---|---|---|---|---|
\((W^{(c)}_{ij}) - (rf^{(c)}_{ij})\) | 0.085 | (4.8 × 10−4) | −0.019 | (4.3 × 10−1) |
\((W^{(s)}_{ij}) - (rf^{(s)}_{ij})\) | 0.50 | (2.4 × 10−79) | 0.030 | (3.1 × 10−1) |
\((W^{(c)}_{ij}) - (rr^{(c)}_{ij})\) | 0.16 | (7.2 × 10−8) | 0.24 | (1.1 × 10−16) |
\((W^{(s)}_{ij}) - (rr^{(s)}_{ij})\) | −0.040 | (2.7 × 10−1) | −0.062 | (8.8 × 10−2) |
\((W^{(c)}_{ij}) - (W^{(cr)}_{ij})\) | 0.16 | (3.8 × 10−8) | 0.23 | (2.5 × 10−15) |
\((W^{(s)}_{ij}) - (W^{(sr)}_{ij})\) | 0.60 | (4.5 × 10−76) | 0.23 | (2.5 × 10−15) |