Abstract:
In recent years, the accelerated escalation of smart phones has led to the increasing popularity of Location-Based Social Networking(LBSN) sites such as Foursquare, Facebook Places, Twitter etc. LBSNs allow and encourage users to publish information about their current location or visiting places through check-ins and offer them to associate their posts and photos with their check-ins and share with their friends and family as well as tagging them. These produce fast growing, ne-grained and vast in volume data and provide a means of user pro ling and modeling. Huge volume of user generated data of social media presents an opportunity to find interesting insights about users' preferences of places at different times. Prediction of users' daily routine, finding users' location preference, identification of users' mobility patterns from the check-in datasets covers the current state of the art. Plethora of works have been done to find such comprehensions about user activity from check-in dataset of social media by considering various aspects such as, frequency of check-ins, time of check-ins, venue of check-ins etc. The knowledge of such intuitions about users' preferences of places or activities has wide range of applications covering social media commerce, targeted advertisement, influencer marketing etc. Among all the works done so far, no one considers the influence of weather on human life while predicting or finding users' mobility or activity pattern, though its effect is enormous. Earlier psychological studies show that weather has a strong influence on human life and its consideration for users' where-abouts and what-abouts prediction is more constructive and pragmatic. Motivating from all these observations we propose the first approach to find user activity and mobility pattern from social media data based on weather condition. In this thesis, we develop several machine learning based models to predict future activity, visiting places and travel mode of users' from previous users' check-ins and travel patterns on a given weather condition, for example, a user may prefer to visit sea beaches on a sunny weather, whereas an indoor entertainment on a rainy weather. Again he or she may prefer cycling on a clear weather whereas taxi or private car on a rainy weather.
