Ez egy előző félévben kiírt, archivált téma.
Nowadays the technology exists to collect large amount and fine granular mobile user and application data, which can be subject of machine learning and automated analysis tools to generate valuable insight. An interesting topic is to understand the behavior of users through the analysis of their mobile phone usage, a technique called user profiling or segmentation. Such insight has become increasingly important for mobile operators for value creation through personalized services, care, marketing and demand based network planning. The main objective of this task is to investigate and validate methods and algorithms developed for user profiling. The algorithms allow the detection of user behavior patterns, temporary or long term behavioral changes and the creation of typical user clusters. The work should start with a survey to collect the technical approaches and mathematical models that are suitable for creating user profiles and identify the meaningful clusters. The scope of the work includes: - validate existing user segmentation algorithms on real or realistic data sets (requires data transformation and/or algorithm adaptation/parameterization/development) - algorithm research for application layer traffic profiling - investigate practical implementation problems (scalability, parallel computation, optimization, etc.) Advantage: - well founded mathematical background on statistics - knowledge of at least one programming or scripting language - familiar with database technologies (relational, no-SQL, distributed key-value pair, etc.) - familiar with Big Data frameworks (Hadoop, Storm, etc.)