3. Knowledge extraction

Methods and tools for automatic extraction of useful knowledge from data have been applied and developed, to support decision-making processes in problems affecting the quality of a multimodal service of Urban Public Transport. For instance, several knowledge discovery techniques have been developed for extracting useful knowledge from data:

– A process mining technique for inferring operational information from the geolocation data. This information can be used to get insight into the underlying business processes.

– A sub-group discovery technique for finding interesting frequent patterns from ticket validations [28]. The behaviour of travellers in the public transport throughout the day is analysed to identify significant sub-sequences of events characterized by statistically unusual aspects of interest.

– A geographic analytical processing tool for assessing the accessibility of public transport based on the walking ability of different groups of the population. Demographic and anthropometric data were used to identify social exclusion in the public transport of Porto Metropolitan Area.

– An artificial neural network model for estimating the demand of urban public transport buses based on weather conditions. Transit bus ridership and weather conditions were linked under the assumption that individuals choose the travel mode based on the weather conditions that are observed at the departure time, or during the hour before and two hours before.