The past year, Parcify collaborated with the University of Antwerp in an innovative data science research project, accurately predicting future handover time and location. This work was the first to address the problem of one-week-ahead customer location prediction, using a unique dataset on mobile location data collected through the Parcify app. By being able to predict where someone will be to receive his or her package, while taking into account privacy, more satisfied customers and more efficient delivery can be obtained.
Parcify is a Belgian start-up company that delivers parcels wherever and whenever the customer wants. Via a mobile application the customer is connected to a local courier that delivers the purchases to the customer's location of choice. This way, Parcify ensures that you never have to miss your parcel again. For the moment, they are operating in Brussels, Antwerp, Ghent and Amsterdam and are partnering with well-known web shops such as Bol.com, Zalando and MediaMarkt. Thanks to the data science collaboration with the Applied Data Mining research group of prof. David Martens, Parcify has now the technology on his side to disrupt the logistics sector and strive for more sustainable and efficient parcel delivery.
Doctoral researcher Stiene Praet from the ADM research group had a fantastic time at Parcify, and already published a paper on the subject: “This is a great example of a successful collaboration that benefits all parties. Parcify can utilize the developed techniques to strengthen their business and at the academic level, new opportunities have been created for future research. Also, personally, I have learned a lot from this collaboration and I really enjoyed the time I spent at the Parcify office.”
Patrick Leysen, CEO of Parcify: "This collaboration has lifted Parcify to a next level in the world of Applied Data Mining. We have been impressed by the cutting-edge technology and research skills of the University of Antwerp.”
Ellen Tobback received the Paragon Award for the Best Paper at the Credit Scoring and Credit Control XV conference at the University of Edinburgh in August 2017 for her paper "Retail credit scoring using fine-grained payment data". The paper, co-authored with her supervisor David Martens, looks at the granular use of debit account transactions. Using a real-life data set of 183 million transactions made by 2.6 million customers. The data shows that amongst the clients that have transacted with at least one defaulter, 52% are defaulters themselves, while the overall default rate is 0.8%. The paper shows that by using intelligent algorithms that use such data to make credit scoring models, already available “big data” assets can be leveraged for improved risk management.
At the conference dinner, the jury awarded the price to Ellen with kind words. Specifically the innovativeness of the type of data was emphasized. The paper was chosen among the 72 papers that were presented. In total more than 400 academics and bankers attended the bi-annual conference in Edinburgh, seen as thé conference in the credit risk domain.