Analysis of user behaviors by mining large network data sets

• We process 16-week-long CDR data of one million users with fuzzy clustering.
• We find some relations between ARPU level and users behavior patterns.
• We prove that the mobility of users is related to ARPU and communication behavior.
• Results indicate that the top ARPU level users are the most “lonely” ones.

Understanding the intelligence of human behaviors by mining petabytes of network data represents the tendency in social behaviors research and shows great significance on Internet application designing and service expansion. Meanwhile, the running mobile networks that generate huge data can be the best social sensor for these studies. This paper investigates a practical case of mobile network aided social sensing which uncovers some features of users’ behaviors in mobile networks by intelligently processing the big data. The paper studies the users’ behaviors with regard to communication, movement, and consumption based on large user data sets. The main contribution of the study is some findings on the relations among these behavior features. We find that the users’ calling behaviors are different despite their monthly expenditures being similar, though different consumption level users may have similar communication behaviors. We also find that statistically users with the higher mobility contribute more ARPU than those with lower mobility. Additionally, we also find that the top consumption level users are the most “lonely” ones by exploring the movement clustering patterns of users. These findings are significant to instruct marketing strategies for telecommunication industry.