Financial Institutions
Insurance Agency Performance & Loss Ratio
How an insurer separated genuine agency underperformance from bad data — and why the retention programme had to stop using loss ratio as a proxy.
View projectIndustry · 02
Risk, credit and customer analytics for lenders and financial services teams — built with explainability, auditability and regulatory review in mind.


Will a borrower fail to pay the bill of their credit card? How much would the lender lose from the default? Prediction analysis for default can give a snapshot of all customers at a specific point in a period with a 'default or not' flag, indicating whether they will default three months later.
Perhaps actions would be taken to reduce the risk of losses.
9 projects
Financial Institutions
How an insurer separated genuine agency underperformance from bad data — and why the retention programme had to stop using loss ratio as a proxy.
View projectFinancial Institutions
How a consumer lender moved collections from post-missed-payment recovery to a calibrated pre-due-date worklist — with the sampling correction done properly and thresholds set by team capacity.
View projectFinancial Institutions
How a retail bank turned a broad outbound list into a pre-call propensity score — call-duration leakage removed, lift measured where the dials are spent, and the result tested against a holdout.
View projectFinancial Institutions
How a credit team replaced a flat checklist with a time-aware triage score — reason codes, cost-weighted thresholds and validation an internal credit committee can defend.
View projectFinancial Institutions
A travel insurance review that separates premium, commission and claims — product and channel performance, exposure-adjusted claim rates and a missing-data pattern that changes what a segment means.
View projectFinancial Institutions
How a card issuer replaced late rules-based review with a three-month default score — observation-point discipline, a transparent scorecard baseline and thresholds tied to collections capacity.
View projectFinancial Institutions
A prepaid wallet churn review where the hardest work was upstream of the model — label polarity, repeated subscriber snapshots, overlapping 30/90-day windows and a retention action the business could actually fund.
View projectFinancial Institutions
A labelled claims file where the obvious signals point the wrong way: witness presence is lower on fraudulent claims, payout size barely differs, and the real constraint is how many claims the investigations team can review.
View projectFinancial Institutions
A 252,000-row application file with no missing values, an income column with no unit and 19,121 records where work experience exceeds what is possible for the applicant's age.
View projectFocus on your business; let us back you up with data support