Credit2B, a leader in automated credit information management solutions, recently announced a further investment in its artificial intelligence capabilities applied to business credit scoring.
The methods employed by credit analysts have not devolved in many years. The old process is to create customer scorecards, assign weights to variables, and over time review how the score cards would have performed using some level of historical experience, and then program those scores. These hard-coded, fixed weight models do not readily adapt to changing conditions, are hard to modify, and very rarely is there confirmation of the validity of the scores. This “Generation 1.0” approach, in use for decades, can now being superseded due to advanced technologies and powerful computing engines.
Credit2B employed data scientists to further enhance its advanced R-ScoresTM system to a Generation 2.0 system soon to be released to make better use of the latest artificial intelligence tools. Quoting Irina Rabinovich, Credit2B CTO, “We have dramatically increased the application of advanced technologies and have research showing that machine learning can take traditional models like the Z-score and enhance their performance very significantly”.
Using artificial intelligence and neural networks with variable multiple sources of live economic and trading data, these complex algorithms are able to adjust weights automatically based on changes in underlying conditions and data. Scores can now utilize vast amounts of diverse data elements, including non-traditional elements and big data.
In Generation 2.0 scoring, Credit2B has refined the methodology so that clients can pick exactly the outcomes they want to predict. For example, are they concerned about bankruptcy risk, or predicting if a company will pay its bills 90 days late. Credit Managers, CFOs, and Controllers will value this approach because each company has its own unique criteria to manage customer risk. Now they are able to pinpoint outcomes with greater accuracy.
Recently, Credit2B published a paper in partnership with the Credit Research Foundation, the leading research organization for trade credit, highlighting many of the underlying methods and how the technique is making credit scores better.
In addition, Credit2B has deepened its partnership with major third party specialists, including the work being done by Professor Anna Costello from the University of Michigan, a subject matter expert in the area of supply chain financing and risk. Professor Costello continues to be a strong catalyst and brings outside thinking to the industry.
As part of this ongoing investment in artificial intelligence, in addition to the debtor-related data, Credit2B will incorporate new exogenous data elements that impact company risk. Examples of this include country, region, macro-economic data, and open source data that would help educate and curate the training models effectively. Quoting Credit2B’s CTO, “Machine Learning models are powerful because they get better over time with more data and information, which is a vast improvement over prior scoring methods”.