Skip to main content

Table 6 Prediction performance on the training set using the Rotation Forest classifier, 10-fold cross validation, with balanced positive and negative samples (10,000 examples)

From: Link creation and information spreading over social and communication ties in an interest-based online social network

Features

FP rate

FN rate

Accuracy

F-value

AUC

Profile

0.279

0.364

0.679

0.678

0.741

Structural

0.241

0.298

0.730

0.730

0.805

All

0.223

0.264

0.757

0.757

0.835

Restricted

0.219

0.279

0.751

0.751

0.826

  1. Four different combinations of features are considered, the “Restricted” category includes a smaller sets of features designated as more predictive by the feature selection process.