Skip to main content

Table 4 Regression results for the tie persistence using generalized linear models for the two prediction models in the Facebook dataset

From: Temporal patterns behind the strength of persistent ties

Feature

Model 1

Model 2

Full

Simplified

Full

Simplified

\(w_{ij}\)

0.839

1.228

1.709

2.041

(0.044)

(0.066)

(0.180)

(0.105)

\(r_{ij}\)

0.118

 

0.470

 

(0.034)

 

(0.111)

 

\(o_{ij}\)

0.269

 

0.384

 

(0.035)

 

(0.111)

 

\(k_{ij}\)

−0.008

 

−0.231

 

(0.031)

 

(0.099)

 

\(\mu _{ij}^{\mathrm{int}}\)

−0.0139

 

0.006

 

(0.030)

 

(0.095)

 

\(\hat{f}_{ij}\)

−0.608

−0.306

  

(0.0390)

(0.016)

  

\(t_{ij}^{\mathrm{min}}\)

−0.329

 

−0.402

 

(0.039)

 

(0.128)

 

\(\mathit{cv}_{ij}\)

−0.286

 

−0.229

−2.394

(0.037)

 

(0.103)

(0.224)

\(\mu _{ij}^{\mathrm{chats}}\)

0.024

 

−0.047

 

(0.035)

 

(0.114)

 

\(a_{ij}\)

0.122

 

0.115

 

(0.036)

 

(0.111)

 

Constant

0.435

1.951

0.686

−4.525

(0.032)

(0.155)

(0.115)

(0.265)

Number of observations

5466

5466

667

667

Performance

Model 1

Model 2

Full

Simplified

Full

Simplified

Accuracy

0.690

0.688

0.798

0.799

Sensitivity

0.770

0.780

0.814

0.802

Specificity

0.583

0.567

0.776

0.797

  1. Coefficients are shown with uncertainties (standard errors) in parentheses. Model Full include all the features described in the text, while model Simplified only includes the most important two features. Note: p<0.1; p<0.05; p<0.01.