Sketch of User Engagement life cycle. The red arrows indicate the possible places of interaction with technology.
We explore three different frameworks: (i) survival analysis models, (ii) regression models, and (iii) random forest models. In doing so, we aim to understand if (and when) a user will disengage after a given period of time. One of the most important steps to achieve better results with these type of models is to group users based on their past behaviour. In that way, we are able to separate – or “cluster” – users based on how (often) they interact with the mobile application.
Pearson-gamma correlation for the optimized (blue) and standard (red) clustering procedures against the number of clusters.
ROC curves for the RF model. The test set curve is shown in red, followed by the Validation 1,2,
and 3 sets in green, blue, and cyan, respectively.
Modelling User Engagement in Mobile Applications
We propose a simple numerical framework to model User Engagement in mobile applications.
Role |
Data Scientist |
For |
Data Science: Methods, Infrastructure, and Applications Journal |
Type |
Peer-reviewed scientific journal |
URL |
datasciencehub.net/system/files/ds-paper-608.pdf |
Language And System |
R on AWS |
Together With |
E. Grua, I. Malavolta, M. Starcevic, E. Weusthof, J .van den Hoven |