Cosmos Scoring

Predictive analytic solutions often sit between data-driven strategy trees and collection optimization in terms of complexity, and we harness our custom in- house scorecard which is built on multi linear & logistic regression models using your own collection data.

Our scorecards offer much finer granularity and higher predictive capabilities because the scores relate to a probability or odds of an event happening. The scorecard is derived from algorithms that identify the most predictive characteristics from within our data sources and consider complex interactions so you can use a combination of sets to make an accurate prediction on future outcomes.
The scorecard itself consists of 8 to 12 characteristics, each of which has a range of values. Associated with each characteristic attribute or range is a partial score. You must add up one partial score from each characteristic to determine a final score. It would be difficult to assess all the variables and work out these computations using trees alone.

Connections can be found between data available at the beginning of the collection campaign and the favourable outcomes achieved at the end of treatment cycles, such as right-party contact, contactability, repayment, proportion of debt repaid and more.

The visual shows some of the touchpoints where our predictive models can support your collection objectives.