Development and deployment of a prediction algorithm in the management of credit files.
The prediction is based on the identification of many elements that independently seem unreviewable but cumulative become factors of complexity.
This identification allows better management of the files in terms of allocated resources and responses.
Implementation of the algorithm and deployment in a SparkMLlib environment.
Made available through a data vizualization solution widely used in organizations (eg Tableau).
Credits: Corp Events – Big Data Paris 2019
Mikaël Le Bars (Natixis), Florian Caringi (Natixis), Nicolas Korchia (Indexima)