A Time-Dependent Principal Components-Based Dimension Reduction Approach to Analyzing the Influence of Product Interventions on User Engagement with Mobile Applications

We propose a new framework of identifying mobile applications interventions (e.g., updates, new features, or new versions) with positive (or negative) users' attitudes, by analyzing temporal changes in a defined set of usage metrics, yielding a general metric, a Mobile Application User Engagement (MAUE). The new metric is a linear combination of user engagement time-series metrics, accounting for the largest amount of the variance in usage data via principal component analysis (PCA). Our proposed approach has been applied to 1533651 behavioral data records of The Weather Company (TWC) IOS users, to analyze the influence of an app update that occurred during the time interval of data collection. Our results indicate that the time-dependent fluctuations of the MAUE trends for the epochs before and after the app update are characterized with a power-law decrease, where a faster decrease is observed for the time period after app update and indicates a higher MAUE score for this time period.