An exploratory analysis on user Behavior regularity in the mobile Internet

The ever-changing nature of the mobile Internet contributes to the difficulties encountered when experts try to identify the user behavior characteristics. Using thin channels with so-called 24-hour 365-day always on nature, it is crucial to understand regularity of user access in the mobile Internet. It is leveraged by the mobile Internet-specific features like user identifies provided by wireless carriers. The author attempts to identify the easy-gone mobile Internet users from regularity dimension using a long-term user log with user identifiers. The author proposes an interval probability comparison method to predict the user behavior in the next month. The experiment from the mobile clickstream data shows the positive effect of the proposed method.