Machine Learning Pressure Broadening Parameters for Exoplanetary Studies

Here is the data and trained model used to produce the multi-perturber machine learning .broad files available on the ExoMol website. This work is described in full detail in my PhD thesis: "Machine Learning Pressure Broadening Parameters for Exoplanetary Studies", and also in an upcoming paper to follow on from 'Predicting the Rotational Dependence of Line Broadening using Machine Learning'. "data_train.csv" is a large csv of compiled broadening data with many active and perturbing species, from many sources, described fully in the paper and thesis. It is the data used to train the the ML model in "production_model.joblib". This model can now be used to predict broadening data for unseen active-perturbing species pairs - and has been used to produce the .broad files available on exomol.com. "production_features.csv" and "production_metadata.json" are what their names suggest. The code used to produce this data file and ML model will be available on github: https://github.com/erg43/Machine-Leaning-Pressure-Broadening.