Scalar coupling, or J coupling constants contain information about relative bond distances and angles in a molecule. This is useful for determining the structure of molecules. We used a dataset containing over 4.6M observations of molecular properties to build a model that can predict the scalar coupling constants given the types of atoms, their indices on the molecule, their coordinate positions in space, and their dipole moments.
This github repo contains the code to recreate visualizations and models for this project.
Medium Article https://medium.com/@liztersahakyan/predicting-scalar-coupling-constants-using-machine-learning-c213af14e862
Kaggle Competition Link https://www.kaggle.com/c/champs-scalar-coupling