Skip to main content
Log in

Table 5 Model fit indices of the stationary, linear and breakpoint model for each of the four forum-level outcome measures

From: The temporal evolution of a far-right forum

Outcome measure

Model

Evaluation metric

AIC

BIC

MAE

RMSE

Forum activity (number of posts)

Stationary model

3088.85

3094.98

3207.95

4201.70

Linear model

3029.02

3038.39

2821.48

3456.99

Breakpoint model (5)

2814.08

2835.25

1278.11

1706.29

Number of words per post

Stationary model

1045.71

1051.84

5.37

6.54

Linear model

994.55

1003.74

4.38

5.53

Breakpoint model (4)

917.95

936.32

3.24

4.25

Proportion posts with extremist language

Stationary model

− 766.05

− 759.92

0.0260

0.0212

Linear model

− 834.57

− 825.38

0.0139

0.0169

Breakpoint model (4)

− 1013.08

− 994.71

0.0073

0.0094

Average extremist language score per post

Stationary model

253.29

259.42

0.4483

0.5326

Linear model

252.95

262.14

0.4357

0.5286

Breakpoint model (3)

115.50

130.81

0.2510

0.3379

  1. The integer in brackets after the breakpoint model indicates the number of regime shifts in the model. The best model fit is highlighted in bold
  2. AIC Akaike’s information criterion, BIC Bayesian information criterion, MEA mean absolute error, RMSE root mean square error
Morty Proxy This is a proxified and sanitized view of the page, visit original site.