Approaches for prioritizing feature improvements extracted from app reviews

App reviews contain valuable feedback about what features should be fixed and improved. This feedback could be 'mined' to facilitate app maintenance and evolution. While requirements are routinely extracted from post-release users' feedback in traditional projects, app reviews are often generated by a much larger client-base with competing needs and priorities and ad hoc structure. Although there has been interest aimed at exploring the nature of issues reported in app reviews (e.g., bugs and enhancement requests), prioritizing these outcomes for improving and evolving apps hasn't received much attention. In this preliminary study we aim to bridge this gap by proposing three prioritization approaches. Driven by literature in other domains, we identify four attributes (frequency, rating, negative emotions and deontics) that serve as the base constructs for prioritization. Thereafter, using these four constructs, we develop three approaches (individual attribute-based approach, weighted approach and regression-based approach) that may help developers to prioritize features for improvements. We evaluate our approaches in constructing multiple prioritized lists of features using reviews from the MyTracks app. It is anticipated that these prioritized lists could allow developers to better focus their efforts in deciding which aspects of their apps to improve.