We are introducing a series of notebooks on machine learning in Python. Notebooks describe a complete machine learning workflow using many examples and explanations. They contain hints about what to do when model performance is not satisfactory. There are also linked resources to study presented topics in more depth.
There are many online resources for machine learning. Very few of them help to solve real production problems, though. We are trying to close this gap, from data exploration to model explanation.
The notebooks solve weather prediction problem in Australia. They cover the following main topics:
- Data preparation, data exploration, feature engineering
- Modeling, evaluating, explaining, getting models into production
- Experimentation using A/B tests
- Error-based machine learning and implementation of simple deep nets in TensorFlow
These notebooks will be useful as a reference for novice machine learning practitioners. But skilled practitioners could find there a few exciting topics as well.
These notebooks originated in the Introduction to Machine Learning with Python course. It was lectured by Avast as a part of the Economics Discovery Hub at CERGE-EI in fall 2019.