/featuredriven-hawkes

Code and Data for CIKM Paper Feature Driven and Point Process Approaches for Popularity Prediction

Primary LanguageJupyter NotebookGNU General Public License v3.0GPL-3.0

Feature Driven and Point Process Approaches for Popularity Prediction

Code and Data for CIKM Paper Feature Driven and Point Process Approaches for Popularity Prediction

We also provide a tutorial for using hawkes process to model real cascades

Steps covered are as follows:

  • Plotting Memory function for the Social Kernel
  • Simmulating events from a Marked Hawkes Process
  • Modeling a retweet cascade with marked Hawkes Process
  • Using the model to make predictions of final size

All this is illustrated with the notebook file provided along with the rscripts.