In the relentless pursuit of competitive advantage, companies have adopted data driven decision making in a big way. A big manifestation of this is the proliferation of predictive models that are being deployed to lift business outcomes in a whole host of applications.
These range from trying to understand customers, segmentation and targeting, propensity to buy, forecasting, etc. etc. A typical firm would use 20 – 400+ models depending on portfolio size, number of products and analytics’ maturity.
Banking and Financial Services is one of those “model guzzler” industries where they are found in every nook and corner. With these models spreading their tentacles within the Banking organizations, monitoring and governance around their usage and applicability is of paramount importance.
In the radically changed post 2008 Dodd-Frank world of Financial Services, regulatory requirements are endeavoring to keep a hawk eye on the performance of models and the resulting implications. The Office of the Comptroller of Currency (OCC) revised and significantly expanded the supervisory guidance contained in the Risk Bulletin on Model Validation, which was co-issued with the Federal Reserve in April 2011. It is somewhat believed that this was in response to observed systemic weaknesses in Model based Risk Management programs at Banks. Therefore, the ability to continually track and monitor performance of these models cannot be over-emphasized. This also needs to go beyond a case-by-case approach of responding to weaknesses in a single model. This is where a structured, integrated and a systematic model governance platform from BRIDGEi2i comes in. Key Features of this platform include:
- Supports a repertoire of statistical models – Be it a Logistic Model implemented to predict likelihood of an event or an ARIMA model to forecast losses or a decision tree to run a marketing campaign, the platform supports all of these. The inbuilt sophisticated algorithms can help track key metrics suited to any of these models.
- Customised Metrics & Charts – Every model is tracked for its Accuracy, Efficiency and Stability. The tool comes pre-built for tracking several standard metrics like Population Stability Index, Gini Coefficient, Mean Weighted Attribute etc. However, if the business user wishes to track alternate metrics, the tool can be customised to do so.
- Manage and Edit Models – If the Business environment prompts you to tweak your model – be it adding or deleting an exclusion criteria, swapping –in or swapping out of a segment or even changing the definition of your target variable, you can make that change with an easy toggle of radio buttons on the tool and the tracking of the revised model is set in motion.
- Ad-hoc & Consistent Reports – The tool’s well-oiled algorithms help generate standard reports at monthly / quarterly intervals. Should the user need any customized reports on an ad-hoc basis the report may be generated with ease.
- Access Controls & Data – Given the nature of business we understand that data security is critical. The access control feature of the tool ensures that model data and other related documents can be accessed and viewed only by the model owner and related stakeholders. Model owner has the ability to set the “Read”- “Write” – “Modify” access levels to select users.
Key Users: The Model Governance Tool may be customized to suit an organisation’s requirements and it comes in-handy for:
- Business Managers, who need pro-active alerts to aid in quick and timely decision making. It gives managers a pan-organization view of all models implemented. It aids managers to quickly respond to requests from the regulators and also be compliant with regulatory reporting requirements.
- Analysts, who are keen to know and are required to keep a constant tab the performance of models developed. The tool also provides insights on aspects and attributes where a particular model needs tweaking.
The ‘Model Governance’ tool from BRIDGEi2i, is an analyst friendly and easy to implement enterprise server tool. This tool simplifies the entire lifecycle of model management – From the point the model goes LIVE to regular tracking on model performance and audits on model enhancement. The tool besides serving as a platform for pro-active alerts on model performance also serves as a repository on model documentation and model approvals. Most importantly it brings enterprise wide visibility to models and consistent objective way of measuring their performance, giving regulators comfort around model governance.
The author, Nandhini Giridharan is an Analytics Consultant at BRIDGEi2i, a company on a mission to unleash the power of analytics and transform the lives of enterprises and individuals alike. We believe that the solutions to almost all intractable problems lies in our ability to mine & contextualize data. We want to help organizations prepare for adversity of all types by embracing data driven calculated risk taking, as a way of life. To know more visit www.bridgei2i.com
The views and opinions expressed in this article are those of the author and do not necessarily reflect the official position or viewpoint of BRIDGEi2i.