Pattern recognition in the context of timetable deviations in the public transport system of Berliner Verkehrsbetriebe Gesellschaft (BVG) for operational optimization

Context

Punctuality is an important quality criterion in local public transport. It is defined as the correspondence between the planned arrival and departure times (see timetable) and the actual arrival and departure times. A distinction is made between early arrivals and delays.

The basis for the thesis is data on punctuality from an operational and customer perspective for the three operating branches of the BVG (subway, streetcar, and bus), as well as other relevant traffic and infrastructure data. As part of the thesis, this data is to be processed in the first step, i.e., outliers are to be identified, and any missing values are to be imputed.

The next step involves a thorough evaluation and analysis of the data, using machine learning approaches to identify patterns and form suitable clusters. Meaningful visualizations of the results are then created based on the findings in order to present the results in a concise manner. Based on the findings, strategies will then be developed to optimize public transport operations in Berlin.

The thesis is announced by the FZI office in Berlin and the BVG in Berlin, therefore working in Berlin (temporarily) is a requirement. 

Goals

Improvement of punctuality for the three operating branches (subway, streetcar, bus) of BVG: 

• Concept for the general processing of punctuality data.
• Analysis model for operational and infrastructure data including the integration of other relevant data sources, such as weather data, passenger volumes and staff availability.
• Learning processes for pattern recognition in the context of operational irregularities such as early arrivals and delays.
• Visualization and documentation of the analysis results obtained.
• Strategic model for the sustainable improvement of public transport operations and the punctuality of means of transport.

Requirements

• You are studying for a Master's degree in electrical engineering, computer science or a related course.
• You have an affinity for public transportation.
• You have experience in data analysis and machine learning methods.
• You have knowledge of programming (Python, R).
• You have above-average extent of motivation and personal interest as well as a thorough, precise and independent working spirit.
• You have good oral and written German and/or English skills.