Big Data Scoring has completed the first ever generic European credit scorecard based only on social media data. The final product uses Facebook data to evaluate the payment behavior of European consumers. Facebook was chosen as the main source of social media data, due to its market penetration and richness of available data.
To create our model we used data from different European countries. The data was collected in co-operation with our partners with the purpose of credit scoring. The consent of end users was requested and received for the collection and use of data.
An average Facebook account holds 7 000 lines of data (data points) that, we pre-processed and ran through our algorithms to create relevant variables. We then proceeded to build the scorecard by combining the social media data with real-life payment behavior data.
Based on our findings we can proudly say that we have a product, which can be used either individually or together with existing credit scorecards. We offer our social media scorecard as a service to all creditors who wish to take advantage of big data and social media in their underwriting decisions.
How we did it?
Only data from Facebook profiles was used to create the model, which we combined with customer’s real-life payment data. Data was available about the customer’s age, sex, education, activities, videos, hobbies, and much more. Data collection was done in several European countries, which we believe to be different enough socially and economically, to prove our social media scorecard can be used in every European country.
Information on customers´ friends was available through Facebook. We found this data to be just as rich as all other data in Facebook, yet in need of more pre-processing.
In building the social media scorecard, we used exactly the same techniques that are used in creating traditional application scorecards. Extensive work was done on processing the variables, creating new variables and grouping them. As a result we have a proven social media scorecard with gini co-efficient of 0.340.
If your organization wishes to benefit from adding Facebook data to it´s underwriting process today, please do not hesitate to contact us at email@example.com.