“Ensuring the deployed model is delivering the right results”
Ongoing Sustenance is an activity that evaluates if the model continues to give the desired benefit, as ascertained by the business problem at hand, within acceptable limits of accuracy despite changes in macroeconomic environment, industry landscape or business strategy — or if it requires recalibration, redevelopment or complete replacement.
The model monitoring phase begins after a model has been implemented for actual business use. The monitoring process should have a frequency appropriate to the nature of the model, the availability of new data and the magnitude of the risks involved.
In recent times government agencies have mandated that any decisioning based on empirical models be thoroughly vetted at a prescribe frequency to mitigate potential business risks arising out of an inaccurate model in production.
Approach and Methodology
PrADS uses a comprehensive approach and methodology that involves business validation, data integrity audit and statistical verification to ensure that the model continues to give accurate results. For a successful CMM program, the key is to ensure uninterrupted supply of incremental data without excessive expenditure of IT resources. The incremental data could have additional variables with high predictive power in addition to all the variables used by the original model. As part of our CMM deliverables that would validate the business and statistical efficacy, a comprehensive report is presented recommending any minor / major changes.
Post production process to ensure model is still fit for use