Code for Kaggle Competition - Nomad 2018 Predicting Transparent Semiconductors
This is the code for the Kaggle Competition.
I've split the problem into 2 response variables:
- Formation Energy
- Bandgap Energy.
My approach is to model each response separately and combine the results.
I've tried:
- Several Regression methods such as Linear Regression, Support Vector Machines and Elastic Net
- Ensemble Methods (AdaBoost / Gradient Boosting Machines)
- K-Nearest Neighbors
- XGBoost for Bandgap Energy
- Support Vector Regression