This is the repository I have worked with during the hackathon #hack4bestcx at the Deutsche Telekom. The task was to predict customer satisfaction from a dataset including both technical telecommunication data and the contract types of the customers. Given the high dimensionality of the data, I have focused mainly on feature engineering and implemented two different algortihms to find the most helpful features for a prediction: simulated annealing and relief. The data and the results are censored due to the non-disclosure agreement.
https://github.com/PadraigC/FeatSelTutorial.git
https://github.com/kennethleungty/Simulated-Annealing-Feature-Selection.git