Case Study: a major financial institution improves model accuracy with TigerGraph

Major Financial Institution improves model accuracy from 60% to 90% with TigerGraph

The customer, a global bank's wealth management division, faced the challenge of detecting advisor engagement risk within a massive communication network. Their predictive signals were hidden within the relationships between 795 million historical messages and approximately 5 million new daily records, which could not be analyzed using traditional methods. They partnered with vendor TigerGraph to leverage its graph platform to uncover these hidden network patterns.

TigerGraph's solution involved building an AWS Glue pipeline to transform the data into a knowledge graph. This allowed for the computation of advanced graph data science features like PageRank and Local Clustering Coefficient. The implementation by TigerGraph significantly increased the model's accuracy from 60% to 90%, successfully operationalizing daily graph feature generation across the entire massive and growing dataset.