Predictive Scoring Genpact’s Predictive Scoring solutions help businesses make informed, data-driven decisions about their operations. We have over 200 Predictive Scoring experts, most with advanced degrees in Statistics, Mathematics and Econometrics. Besides scorecard development, we provide upstream and downstream services, from data planning and scorecard validation, to implementation, and performance tracking. We also have expertise in using scores with other decision elements to build rule-sets and strategies for scorecard implementation across the Marketing, Risk and Collections functions. Our Approach We leverage internal performance data (application, billing, transaction, financial, etc.) and external data from credit bureaus and other database vendors to build predictive scorecards. Our analysts use a variety of tools (SAS, Model Builder, ASA and Knowledge Seeker, etc). Depending on requirements, advanced techniques such as Logistic Regression, Linear Regression, Survival Analysis, Neural Networks, Chi Square Automated Interaction Detector (CHAID) and Classification & Regression Tree (CART) are applied. We have a robust methodology for data preparation (variable distributions, missing and extreme value treatments, applying exclusions), a pre-developed library of macros and routines for faster deployment, and templates for model validation. We also follow a Six Sigma framework that includes rigorous quality checks. Client Successes For a large consumer lending business: Developed a direct-mail scorecard based on demographics and credit bureau data that increased response six-fold and saved $1 million annually in marketing costs. For a small-ticket commercial lender: Created an acquisition-risk scorecard using data from multiple credit bureaus, cutting loss exposure by $14 million and increasing sales by $333 million (through higher approval rates).