Last weekend we took part in the WHAT datathon – a data science competition accompanying the WHAT behavioral data conference organized by Wakoopa.
The challenge was to uncover business insights in clickstream data. We got data of one month in 2016 of a Dutch panel that shows how people have been using their connected devices (pc’s, tablets and smartphones). This includes which websites, search terms and apps they have been using, at what time and for how long.
Our winning solution was an interactive dashboard in which we visualized fashion shoppers’ journeys across different shops together with many demographic variables available in the Wakoopa’s dataset. We have used a hierarchical clustering algorithm with a custom distance function to split the journeys into 9 distinctive clusters:
Read more on our Rax blog