Kontextsensitives Predictive Process Monitoring: Verbesserung und Implementierung eines Dynamischen Bayes'schen Netz-Ansatzes
Are you interested in a modern, practical, yet theoretically based thesis? Then read on. Customer movements in ski resorts represent logistical processes, which describe the flow of objects (e.g., persons or resources) within networks through time and location. To date there exist no good prediction systems for customer movements in large tourism industries, such as ski resorts, amusement parks or public transport. One promising approach to predict these processes on a large scale is the RegPFA, which is based on a Hidden Markov Model. The goal of this thesis is to extend the existing RegPFA implementation to also consider contextual information (e.g. weather or machine state) and run on high performance computing machines (e.g. Open MPI or Apache Spark). For the implementation, it is helpful if the student has already some proficiency in or the willingness to learn Java and Apache Spark or MPI/Open MPI. If you are interested in the topic, please do not hesitate to contact me.