User Behavior Patterns and Activity Recognition Based on Mobile Big Data

Abstract

User behavior patterns and activity recognition have a wide range of applications in location-based information mining, such as early warning systems, traffic flow planning, urban computing, mobile marketing, social networking, and user portraits. warning systems, traffic flow planning, urban computing, mobile marketing, social networking, and user portraits. Most of the existing related researches is based on GPS information of high-precision mobile phones, but GPS information is not easily accessible and has a small range. Compared to GPS data, base station data includes a wide range of user data and can be collected passively, which has great potential for excavation. Therefore, in this paper, we base on the base station data to identify user activities and mining behavior patterns. To explore the user's traffic pattern, this paper first gives a candidate matching trajectory with traffic mode attributes combined with data such as road network and then proposes the spatio-temporal dynamic programming similarity algorithm to obtain a candidate matching trajectory with the highest similarity to the original trajectory, thus obtaining the trajectory of traffic Methods and analysis of tourist traffic patterns; in the activity identification part, this paper also applies the Hidden Markov Model to model the user's stay area, applies the maximum expectation algorithm to solve the model parameters, and finally uses the Viterbi decoding algorithm to get user activity labels. and analyze the results of the activity recognition.