Advances in Digital Marketing: Social Networks Forecasting by Link Prediction

Abstract: As social networks grow; large systems tend to form a network between thousands of networks. Predicting links has become an emerging business technique for analysis. This study presents the prediction of link and its complexities using ten different predictive techniques. We have examined seven datasets from different fields. The study used machine learning algorithms to see which method delivers more accurate results; moreover, Area under the Curve (AUC) - ROC curve is used to see the performance statistic for classification at various threshold levels. The results show that Random Forest outperforms the K closest neighbor method in terms of accuracy. Keywords: Social Networks, Prediction Techniques, Machine Learning JEL Classification Number: F31, F4