Design And Implementation Of WiFi User Behavior Analysis System Based On K-means Algorithm

The analysis of user behavior has always been a research hotspot.For example,studying the behavior of students and teachers on campus and understanding the operation of various places on campus can provide reference for managers' decision-making;studying the behavior of customers in shopping malls helps to understand the distribution of passenger traffic in the mall and provide customers with personalized recommendations and good commercial value.With the popularization of mobile intelligent terminal devices,more and more users will open the wireless network function of mobile devices and other terminal devices for a long time.As a result,the wireless network probes can conveniently collect a large amount of user behavior data.Because the data collected by the WIFI probe produces a huge amount of data and the label cannot be directly labeled,so the labeling cost is high.It is important to explore an unsupervised algorithm for WIFI probe data mining.This paper studies the design and implementation of WIFI user behavior analysis system based on k-means algorithm.The main research idea includes the following three points:(1)Based on the business scenario of WIFI user behavior analysis,the paper analyzes the system requirements and designs the WIFI user behavior analysis system.The whole system is divided into three parts:data access module,data processing module and data application module.The data access module synchronizes the WIFI probe data to the HDFS to achieve large data volume access and improve fault tolerance.The data processing module performs preprocessing and feature engineering on the original WIFI probe data,and performs data mining from multiple angles.The data application module visually presents the results of the user behavior analysis.(2)This paper constructs a mathematical model of multi-dimensional analysis of WIFI user behavior.The clustering model is constructed from different angles such as users,places and time.Clustering by user can analyze user's behavior habits and activity;clustering by time can analyze the working habits of users in different time periods;clustering by location can be beneficial to control traffic distribution of each location.(3)Based on the requirements of fast and real-time processing of WIFI big data processing and the analysis of the limitations of traditional k-means algorithm,the k-means algorithm used in the data processing module is improved from multiple angles.As a result,it can process real?time WIFI probe data.The paper not only proves the effectiveness of the improved algorithm through experiments,but also applies it to the WIFI user behavior analysis system,which has certain practical significance.