MathWorks Model Inventory

This is the MathWorks model inventory, an interactive environment that allows you to manage all model assets and automate workflows. Our model inventory has multiple stakeholder views at varying levels of granularity from top level views of aggregated statistics showing the impact to the bottom line all the way down to the model of a, view where you have information about the metadata of the model itself. On the right hand side, the dependency view allows you to focus on the interconnection between models and data.

It enables us to quickly find and filter down models of interest. For example, if we care about assessing the impact of modifying the credit card PD model, we can focus on just the relevant models and eliminate the noise. The visual tools and point and click operations make it a snap to create and manage these dependencies.

On the left, there's the ledger, a top-down view that aggregates all models for specific business and product lines. We can immediately identify any areas performing suboptimally and, going one step further, even layer on additional search filters, which is a very powerful way to quickly identify key models of interest. Let's say we care about this provisioning model.

The model panel on the right allows us to easily edit the model details based on our permissions and role. We can even go so far as fully customize the fields themselves and add or remove ones that we want. Since the model inventory provides a single location for all model assets, you no longer have to sign into different systems to find your reports, model documents, or model scripts.

Everything is centralized in one location and can be run from a web browser so you can access it anywhere at any time. And the solution is robust. So you could even manage versions of individual models and manage models written in any language.

Furthermore, the workflow is customizable and can even automate tasks to reduce cost and ensure reproducibility. Automate any type of action-- from executing batch jobs, building model documentation, to setting up testing environments. As an example, here we're triggering a custom action that pre-populates an issue template.

And once submitted, the table of issues is updated in real time. Because findings and exceptions are also managed here, we know exactly how every model is behaving over time with this detailed history-- things such as, do we find the key item of interest, has the problem been remediated, can we still use the model in production, and so forth. And even the documents can be managed here as the inventory integrates with your existing systems.

For example, here, we're showing documents being served up existing SharePoint infrastructure. The tight interoperability means users only need to sign in once to access everything model-related. All this information is available for every model, which is conveniently listed in the Models table on the left.

This view provides the lowest level of granularity, where you can see all aspects of the model metadata, such as materiality, business owner, and exposure. Speaking of exposure, there's a built-in model risk scoring, which is fully customizable. While the example here is more of a traditional risk questionnaire, we can also leverage cutting edge AI, NLP, deep learning, and more.

Then once the score is calculated, all changes can be immediately synced back into the model inventory to ensure consistency. And with all of this information, one of the most important tasks is to make sense of the data. The dashboard is an executive level view into your model ecosystem.

Customizable key risk indicators not only have drill down capabilities but can also run custom algorithms in real time, meaning that you can even use the dashboard to do what if analysis and stress testing. Finally, and perhaps most importantly, every action is being recorded, meaning there's a full audit trail of every change being made. We can go back to any point in time, see the model state and behavior, what were the inputs or outputs, or even rerun that model.

And this is the MathWorks advantage-- complete traceability combined with automation. Learn more today about how we can get your models to market faster while also reducing your costs.