Big data can predict bankruptcies

In the case of a series of replacements of a company's board, management and auditors, or if a company changes sector several times over a short period of time, it could be an indication of problems with the economy or an attempt to circumvent the rules.

Professor at Aarhus University, Kaj Grønbæk and PhD student, Andreas Mathisen are responsible for the development of a new analysis tool that can be used to identify companies at risk of bankruptcy.

The researchers have identified a sequence of events that characterize companies that have gone bankrupt over a ten-year period. With the Danish Business Authority, they have prepared a prototype based on information that is available such as open data in the Business Registration number registry (CVR). The CVR register contains data about companies such as start date, change of branch, auditor or director and new board members.

The researchers have just presented their new tool at a scientific conference in Arizona, USA.

Converts data to images

Based on historical data, the researchers have developed a prototype that can visualize the risk of bankruptcy in white, grayscale and black, of which black indicates the greatest risk.

- When the employees from the Danish Business Authority search for companies facing bankruptcy with this tool, they can for example select a sector and an analysis period from a menu. Afterwards, the tool will convert event data from the selected companies into a visualization, which allows the employee to select companies marked with the black-colored sequence of events, Kaj Grønbæk explains.

Read the article (In Danish).

Photo: Bon Adriel Aseniero, Calgary, Canada