Skip to content

This project was developed during the course Laboratory of Computational Physics

Notifications You must be signed in to change notification settings

bhroben/Feature-importance-methods-of-simulated-binary-black-holes

Repository files navigation

Feature-importance-methods-of-simulated-binary-black-holes

In our work we studied almost 2 million simulations of binary stellar systems, carried out with the rapid-synthesis binary population code MOBSE, searching for the most important features that determine the eventual evolution of a binary stellar system into a merging BBH, source of GWs.

The Goal of the project is to find what features have the highest impact on the evolution of a binary system into a Binary Black Hole. To do this we implemented four methods:

  • Weights of linear SVM
  • Medium Difference in Impurity for RFC
  • Permutation importance
  • Feature dropping

We also used SHAP values. We draw our conclusions in the project presentation.

About

This project was developed during the course Laboratory of Computational Physics

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published