/Energy-Prediction-Semiconductor

Establishment of a model in order to predict formation and bandgap energy of transparent semiconductor.

Primary LanguagePythonMIT LicenseMIT

Predict formation- and bandgap energy in semiconductors

The script was implemented for the purpose of the Nomad2018 challenge run by Kaggle. https://www.kaggle.com/c/nomad2018-predict-transparent-conductors

You can download the required input-files from this location. You can also fork my kernel directly in Kaggle: https://www.kaggle.com/maxkapsecker/starter-feature-engineering-ensemble

Supposed folder structure

alt text

Outcome

Predictions for the formation- and bandgap energies of the testdata.

Results

The RMSLE metric yield currently 0.0549 based on approximately 100 test samples. As soon as the competition ends, I will upload a refined approach to this topic.