Based on the same training data as route optimization, Ruter developed an algorithm that could predict how many people would sneak during a given day, and at which stops the sneaking is most likely to take place. The algorithms are weighted with manual input to ensure that ticket checks do not fall short, for example by not checking more in areas where people generally have a slightly tougher financial life. Analyze complaints and praise on a large scale When the bus is delayed, or something else happens that affects the customer experience, Ruter naturally receives complaints. But feedback about public transport in Oslo does not only come through official channels.

People give their opinions in comment fields, in discussion forums, on Twitter and a whole host of other places around the internet. Among other things, we trained our model with a whole range of different Norwegian and Swedish swear words. Words and phrases I had never even heard before! With Brazil WhatsApp Number List artificial intelligence, they were then able to map which types of feedback were most acute, critical and relevant among other things based on exactly which swear words were used. In this way, they were able to visualize hidden trends in the data, which in turn led to interesting and useful insights.

The most common complaints were about crowding and accessibility for wheelchairs. But one complaint in particular came as a surprise to several: A hidden trend was that among the most common complaints were about the music taste of the drivers, Imam explains, laughing. AI tools have also made it possible for Ruter’s customer service to come up with faster and more relevant responses to complaints. Artificial intelligence has saved around , hours of manual work. This corresponds to approximately one full year man year. Preach crowding on the bus Standing like a herring in a barrel on a packed bus is not very pleasant.