Mantra has delivered quality solutions in the analytics space. Team has deep working background through past experiences:
- Revenue and profit maximization for Financial Services clients – attrition management, credit risk and fraud detection, usage prediction
- Model building & validation and design of experiments
- Portfolio level and CRM Analytics
Leasing Collection Model Validation
Client Profile
The client is a large provider of processing terminal leasing services in the US in the merchant acquisition market
Project Requirement
The scope of work required our team to validate existing models that predicted the efficacy of collections amongst their delinquent leasing customers
The Solution
We designed a three month phase wise approach to collating, organizing and analyzing historical and current transaction level data to measure the effectiveness of the client’s predictive models. This involved creating a Champion Challenger process to provide score evaluation.
Validation Procedure
- Create SAS datasets
- Combine all validation data (over a 6 month time horizon)
- Generate Validation Analysis
Results
Score ranges for Champion and Challenger were created and reports were created based on score distributions for
- Current status
- Collectiveness Analysis
- Bad Detection Analysis
Client Feedback
- The client’s collection models were proven to be accurate
- The client then proceeded to integrate its auto-approval technology platform with the Collections model
- The validation engagement was extended to several other Lines of Business and application areas including activation, auto-approval and fraud detection
Credit Risk Modeling
Client Profile
The client is one of the largest transaction processors in the UK for merchants accepting credit card payments
Project Requirement
The scope of work required our team to develop predictive models in the client defined Mid Risk Management (exposure liability based grouping) subset to predict both credit risk and fraud risk.
The Solution
The solution developed was a comprehensive 5 month engagement that involved several steps including
- Understanding the current MRM process (at a strategic and operational level)
- Data request and pulling of required data (including transaction level and Experian bureau data)
- Retrospective Analysis (including evaluation of current rules, redevelopment of new rules based on applicability and creating new scoring models)
- Design of Experiment (Setting up a test versus control to evaluate the effectiveness of the new process)
Several approaches were evaluated including Naive Bayes, support vector machines, neural networks, Univariate analysis and logistical regression
Results
The project was completed within the estimated timeframe and a comprehensive dashboard and tracking system was developed to continuously monitor the results of the credit and risk prediction models
Client Feedback
Models and tracking system are currently integrated into both the UK and US Lines of Business
|
Employee Login | 
