thaasoph and I are working on this project.
This repository contains jupyter notebooks to predict football games between two nations. The trained algorithm is used to predict the football world cup 2018 in Russia.
We use python and jupyter lab, also we use scikit-learn for the models. The most work is to preprocess the data, using machine learning algorithms takes a much smaller amount of time. To be honest: we decided to do this one month before the cup. Things to do better next time: data preprocessing/preparation and more thinking about the ML-algorithm.
Most data sources are kaggle-datasets about Football games and players, also transfermarkt.de is a really good source.
This is the result of the first prediction:
(images created with kicker.de)
