The AnalyticsIQ team is skilled in crafting risk models designed to work optimally on your data. When you partner with us on a risk project, you’ll be assigned an experienced team led by one of AIQ’s senior managers.
The following outlines the typical process we use in developing your solution:
Initial Design Meeting: An on-site meeting will be held to discuss the following issues related to the project:
- Introduction of project team (AIQ and client)
- Discussion of project goals
- Discussion of project definitions (dependent variable, data to be used, etc.)
- Discussion of systematic opportunities and limitations
- Assignment of responsibilities
- Project Timeline
Decision Document: Following the initial design meeting, AIQ creates a document that encapsulates all decisions made relevant to the project.
Data Collection: AIQ works with you and manages third parties to capture accurate modeling data in a timely manner. We also perform a full audit on the data when it arrives at AIQ.
Segmentation Analysis: AIQ performs an analysis to determine the optimal number of models.
Reject Inferencing: AIQ uses a proprietary method of removing ‘reject bias’ on the development sample.
Model Development: AIQ’s methods are proprietary, however, we strongly encourage the client to be involved during the model development process.
Model Presentation Meeting: AIQ presents its proposed models to the client and any proposed changes to the model can be performed ‘real time’ during that meeting.
Model Finalization and Documentation: The final model is developed by AIQ. Upon client signoff, final project documentation is produced including scorecards, forecasts, validation tables, and adverse action reason logic.
Programming Support: AIQ works closely with your programmers, or your chosen vendor, to implement the scoring system accurately. A system-to-system audit is also performed by AIQ.
Ongoing Strategy Support: Based on our experience and knowledge of the data, AIQ recommends cutoff/pricing strategies. AIQ will also perform one follow-up validation at no charge.
