This is a collection of interactive notebooks covering different aspects of machine learning primarily for chemistry/materials science.
To run the notebooks on your local machine, you will want to set up the correct conda
environment.
conda env create -f ml-mater-conda.yaml
- Classical ML models for generic classification
- classification_decision_tree.ipynb
- Classical ML models for property prediction
- regression_decision_tree.ipynb
- sulfides_exercise.ipynb
- Deep neural networks for classifying spectra
- dnn_for_spectra.ipynb
- Interpretable machine learning
- interpretable_ml.ipynb
- Navigate to http://colab.research.google.com/
- Sign in with your Google account
- On the menu that appears choose Github
- In the searchbar type: https://github.com/keeeto/ml-training-materials
- Choose the notebook you want to open
- For the practical; open
regression_decision_tree_practical.ipynb
- If you want to save and reload the notebooks you will need to save a copy of the original notebook to your Google Drive
- Go to File > Save copy in Drive
- When you want to reload navigate to: http://colab.research.google.com/
- Got to File > Open
- Choose the Google Drive tab, you should now see your saved notebooks.
These notebooks were developed for various projects and courses.
Is largely based on material from the excellent Python Data Science Handbook by Jake VanDer Plas and is material that I have used teaching the SciML Introduction to Machine Learning course.
Is based on material published in the paper Data-Driven Discovery of Photoactive Quaternary Oxides Using First-Principles Machine Learning.
Is an excercise that I developed as part of my course - Machine Learning for Chemists, that I teach at the University of Reading
Was developed as part of the SciML Introduction to Machine Learning course.
Is material to accompany a chapter on Building Trust in Machine Learning in my book Machine Learning in Materials Science.