Description

Prototype that uses machine learning to attempt prediction of mobile phone subscriber downgrades/upgrades from usage patterns.

Contents

• prepare-data: Synthetic datasets used for the prototype. Also contains Groovy code that's used to clean and label the datasets.
• train-model: Machine learning model training and prediction using (primarily) scikit-learn and, to a lesser extent, TensorFlow.

See here for a full description of the prototype and its outcomes.