Table 1 Coefficients and p-values for linear regressions whose regressors correspond to meta-paths of length 2 to explain the user-hasthag distribution (UH). Model A0 corresponds to the null model: no predictor but one intercept that is the average of the explained variable
From: Link weights recovery in heterogeneous information networks
Mod. | Meta-Paths | Coef. | \({\varvec{p}}\) -values | \({\varvec{r}}^\mathbf{2 }\) |
|---|---|---|---|---|
A0 | Average : 1.8704e−05 | 0 | ||
A1 | RT-UH | 0.6273 | – | 0.3594 |
B1 | RP-UH | 0.4291 | – | 0.2289 |
C1 | MT-UH | 1.0289 | – | 0.4606 |
A2 | RT-UH | 0.5795 | 0.0062 | 0.6116 |
RP-UH | 0.3957 | 0.0105 | ||
B2 | RT-UH | \(-\)0.3578 | 0.0612 | 0.5943 |
MT-UH | 1.4534 | 0.0087 | ||
C2 | RP-UH | 0.0051 | 0.0138 | 0.6111 |
MT-UH | 0.9391 | 0.0057 | ||
A3 | RT-UH | \(-\)0.1283 | 0.0791 | 0.6818 |
RP-UH | 0.0791 | 0.0113 | ||
MT-UH | 1.1466 | 0.0111 | ||