| Popis: |
RESUMOIn recent years, the use of mobile applications for digital serviceis being widely deployed in a varied range of contexts. With this,predicting the possibility of churn is vital for selecting users thatcould be targeted with user-retention campaigns. This techniqueis commonly referred as Churn Prediction Problem (CPP). Moststudies in the literature use traditional machine learning techniquesto predict churn, and neglect the users’ privacy. In this work, wepropose a privacy-preserving solution that uses neural network topredict churn of mobile services. Our solution, called AutoRGNN,requires only the installation and uninstallation sequences of mobileapps, and integrates Recurrent and Graph Neural Networks. Incomparison with a traditional baseline approach in a large-scaleand real scenario, AutoRGNN was capable to increase the recalland precision up to 19% and 7%, respectively. |
