When you don’t really know what to do with this huge pile of strategic data

A large urban public transport operator had the opportunity to access years of daily historical data about travel on one of the urban networks it’s in charge of—but it was challenging because of the dataset’s size. How do you enable nonexperts to extract value from it for their day-to-day work? And above all, what could they do in a different way with such a material?

Caroline Goulard demonstrates how to leverage data to rethink the way an industry builds its offering, improve customer experience, and feed its prospective reflection. Caroline focuses on the approach, the methodology, and the results achieved for this leading public transport operator. Join in to learn how to turn a complex and massive dataset into an appealing professional tool based on data visualization to better understand habits and needs of customers in the transport sector.

Topics include:
Creating an attractive tool for users who didn’t ask for it:
The main difficulty of this project was that the client’s need was not precisely expressed. Thanks to numerous users’ interviews, three main uses were identified: enhancing processes to construct offers, supporting decision for new line, and anticipating works on the network and its consequences for users.

Converting data science into business language:
With the help of a research lab, the operator set up a clustering algorithm to analyze the dataset, which posed an added challenge. To use this algorithm and parameter a request, you had to master the R language, and, for the output, you had to be able to understand the new dataset delivered. The tool had to be a mediation with the algorithm to make this layer of intelligence usable by professionals who were not data scientists.

Mobilizing representations known by businesses:
This included a visualization on a calendar to observe patterns in travelers’ behaviors, a load line graph on a timeline to see over- or undercapacity, and cartography to examine transterritorial flows. Those users’ habits had to be mixed with other kinds of visualizations, efficient in terms of cognition and ergonomy.