Development of Segmentation Algorithm for Identifying Lifecycle of Smartphone Applications

Main Article Content

Umesha Perera
Maiko Shigeno
Ushio Sumita

Abstract

The concept of the ‘Product Life Cycle’ has been applied extensively to a broad range of products and services, which stay in use for a long time once they are purchased. Accordingly, the stage of the product cycle could be determined by tracing the number of products sold during the underlying time unit (say day, week, month or year) along the time axis. For rapidly growing markets of digital products and services offered through the Internet, however, the underlying product lifecycle may not be figured in a similar manner because such products and services may be installed and uninstalled dynamically over time in a repeated manner. Because of this, combined with the difficulty of acquiring real data, little literature exists for the study of the product lifecycle of the Internet related digital products and services. Recently, a Japanese software development company named Fuller, Inc. developed Android smartphone applications for example, ‘Mr. Mobile, the battery saver,’ which enabled the company to obtain actual data concerning instalments and un-instalments of smartphone applications via the legitimate agreement between the company and the users of these applications. Based on the large volume of actual smartphone application usage data acquired from Fuller, Inc., the purpose of this paper is to fill the above gap by capturing the product lifecycle of smartphone applications. Keeping the feature of dynamic instalment and un-instalment of smartphone applications in mind, a segmentation algorithm is developed so as to classify smartphone applications into three categories based on the concept of the product lifecycle: Class (A) smartphone applications that are active in lifecycle; Class (B) smartphone applications that are in the tail of the lifecycle; and Class (C) smartphone applications that are already diminishing. Numerical results are presented so as to validate the proposed segmentation algorithm.

Article Details

Section
Articles