Article Preview
TopIntroduction
Mobile phone technology has penetrated every aspect of people’s lives. Smartphones have become essential instruments for fast communication and participation in various online activities. The number of mobile applications has grown exponentially and are used on a daily basis covering a wide range of areas including information search, e-commerce, games, entertainment, healthcare, finance, etc. Smartphones have several embedded sensors such as GPS, accelerometer, gyroscope, microphone, camera and Bluetooth. The ubiquity of mobile phones and increasing smartphone usage have generated large amount of behavior related data from the users. The smartphone data, both structured and non-structured data, contain rich amount of information and have potential to provide insights that are extremely useful for different types of businesses such as online retailers, network providers, and application developers.
As more smartphones related data becomes available, smartphone analytics has attracted enormous interests because of its potential business opportunities and impact on various business-related areas. At the same time, many challenges are associated with such data that are characterized by large volume, velocity and variety (3Vs) in understanding and predicting user behavior (Zikopoulos & Eaton, 2011). Marketing field has traditionally relied on market surveys to understand consumer behavior and improve product design. With mobile technologies and big data analytics, customer mobile engagement and strategic marketing decisions can be enhanced by mining smartphone usage data. Taylor and Levin (2014) tested a model by analyzing survey data and reported that more recent the consumer's last visit to the retail store, the larger the effect-size of interest in the app on intention to share information and make a purchase. The power of mobile computing lies in real time communication, delivering what consumer wants and needs at their convenience. Mobile computing can help to engage customers in the key moments of their day, accelerate business process and reach customers with new business services. Mobile apps help users to act in their direct moment of need.
Constructing predictive models of human behavior has been a topic of interest in the area of recommendation systems, context-aware services, and personalized and adaptive interfaces. Lim et al. (2016) used mobile application usage logs provided by a Wi-Fi local area network service provider to characterize temporal behavior of mobile applications and predicted future usages of each mobile application. Papandrea and Giordano (2014) reported based on a study that people tend to move among a limited set of places and that this can be modelled with a user prediction graph, which can further be used to predict the next movement. At the infant stage of mobile computing, some of the important goals for analyzing smartphone usage data include prediction of next behavior, semantic place, next place, and demographic attribute. People behavior associated with smartphone include calling, sending message, surfing online and using an app to fulfill various kinds of needs. Next behavior prediction requires extracting and identifying human behavior patterns and mining serial correlation according to behavior history. Given the multiple locations a smartphone user visits, it is challenging to give semantic meaning to locations in collected data. The prediction of human mobility is relatively difficult, because of limited contextual cues besides spatial-temporal context cues in the mobile phone data set (Laurila et al., 2013). Since it is difficult to obtain private information about users, research efforts have been spent to infer users’ characteristics based upon the contextual information traces.