Uncovering smartphone usage patterns with multi-view mixed membership models

Virtanen, S., Rost, M., Morrison, A., Chalmers, M. and Girolami, M. (2016) Uncovering smartphone usage patterns with multi-view mixed membership models. Stat, 5(1), pp. 57-69. (doi:10.1002/sta4.103)

[img]
Preview
Text
115571.pdf - Published Version
Available under License Creative Commons Attribution.

926kB

Abstract

We present a novel class of mixed membership models for combining information from multiple data sources inferring inter-view and intra-view statistical associations. An important contemporary application of this work is the meaningful synthesis of data sources corresponding to smartphone application usage, app developers' descriptions and customer feedback. We demonstrate the ability of the model to infer meaningful, interpretable and informative app usage patterns based on the app usage data augmented with rich text data describing the apps. We provide quantitative model evaluations showing the model provides significantly better predictive ability than comparative related existing methods.

Item Type:Articles
Status:Published
Refereed:Yes
Glasgow Author(s) Enlighten ID:Chalmers, Professor Matthew and Girolami, Prof Mark and Rost, Dr Mattias and Morrison, Dr Alistair
Authors: Virtanen, S., Rost, M., Morrison, A., Chalmers, M., and Girolami, M.
College/School:College of Science and Engineering > School of Computing Science
Journal Name:Stat
Publisher:John Wiley & Sons, Ltd.
ISSN:2049-1573
ISSN (Online):2049-1573
Copyright Holders:Copyright © 2016 The Authors
First Published:First published in Stat 5(1):57-69
Publisher Policy:Reproduced under a Creative Commons License

University Staff: Request a correction | Enlighten Editors: Update this record

Project CodeAward NoProject NamePrincipal InvestigatorFunder's NameFunder RefLead Dept
566711A population approach to ubicomp system designMatthew ChalmersEngineering & Physical Sciences Research Council (EPSRC)EP/J007617/1COM - COMPUTING SCIENCE