Client
Asset management Fintech
Tech stack
Google Cloud
Solution
Self-service analytics
Service
Data + Analytics
Self-Serve Analytics for an Asset Management Fintech
Our client, a Series D-funded fintech, wanted to carry out Population Stability Index (PSI) analysis dynamically across their loan applicants. Their processes were manual and inflexible, giving limited insight when population shifts were identified. Using Looker, Datatonic’s in-house team of experts built a self-serve tool that analyses PSI dynamically across ever-changing populations of loan applicants.
Our impact
- Created a self-serve PSI tool with LookML, enabling the data science team to create populations dynamically to measure PSI
- Enabled our client to drill down into a range of additional KPI’s for the loan applicants to understand why the PSI had changed
- Greatly reduced data preparation phase and time spent performing PSI analysis
The client
Our client is a fast-growing fintech that brings more efficient credit evaluations and more accurate assessments through deep learning insights. Focusing on fixed income and alternative credit, they offer a variety of discretionary funds to institutional investors (including pension funds and sovereign wealth funds), insurance companies and banks, processing over 90 transactions per minute and handling to-date $4.5b in partnership capital.
The challenge
Our client wanted to carry out Population Stability Index (PSI) analysis dynamically across their loan applicants. PSI is a common analysis for financial services, to measure if the distribution of credit scores have changed over time periods. A high PSI alerts the business to a change in the characteristics of a population, therefore their credit decisioning model would require updating (or the business takes on additional risk).
There were several blockages for the client’s data science team to get this analysis:
- Data Scientists were expected to run SQL queries manually
- Inflexible Process – altering the populations required moderations to the SQL query
- Minimum Alerting when there were population changes
- Lack of ability to drill down when a population shift identified
Our solution
Through Looker, we built a self-serve tool that analysed PSI dynamically across ever-changing populations of loan applicants, with the ability to drill down across a range of variables/KPIs and sped time to insights significantly for the DS team.
