Predicting online video engagement using clickstreams

Abstract

In the nascent days of e-content delivery, having a superior product was enough to give companies an edge against the competition. With today's fiercely competitive market, one needs to be multiple steps ahead, especially when it comes to understanding consumers. Focusing on a large set of web portals owned and managed by a private communications company, we propose methods by which these sites' clickstream data can be used to provide a deep understanding of their visitors, as well as their interests and preferences. We further expand the use of this data to show that it can be effectively used to predict user engagement to video streams.

Predicting User Quitting Ratio in Adaptive Bitrate Video Streaming

“…Moreover, frame rate reduction and stalling have different visibilities and also result in different magnitudes of abandonment [37]. In addition, the interest of users in content results in different actions [37] and tolerance towards technical issues [3], [38]. This can be seen as certain types of content such as animation can have a longer viewing time than other types of contents [37], [38].…”

Anomaly detection of event sequences using multiple temporal resolutions and Markov chains

“…Further, the authors suggested how approaches could be adapted for similar problems. Anomaly detection has been applied by, e.g., Aguiar et al [1] and Meng et al [19]. The former was able to predict early exits of users in a video services, based on the clickstream sequences [1].…”

“…Anomaly detection has been applied by, e.g., Aguiar et al [1] and Meng et al [19]. The former was able to predict early exits of users in a video services, based on the clickstream sequences [1]. This provides deeper understanding of the user-base, e.g., when a user is likely to stop watching specific videos.…”

“…The list of EventIds as such can be considered an event sequence. The event sequence is highly related to the concept of clickstreams, that is, a record of how end-users have interacted with a service [1]. In this case, the interaction is not limited to the explicit actions by the end-user, but also includes interactions initiated by clients' applications.…”

Toward Cost-Effective Mobile Video Streaming through Environment-Aware Watching State Prediction

“…By analyzing the relationship between the time span of video viewing and the power consumption of mobile devices, Li et al [12] adjusted when and how data were downloaded according to the viewing timespan to reduce power consumption. Aguiar [15], Brinton [16,17], and Sinha [18] classified user participation and learning mode by analyzing the click stream. Yang et al [19] used the user’s click stream in a Markov model to predict the user’s next click.…”