AUTOMATION IN RISK MODEL DEVELOPMENT AND DEPLOYMENT
R&C: Could you provide an overview of the current trend of leveraging automation in risk model development and deployment?
Taylor: Automation receives increasing attention these days. Some of this comes from general excitement around artificial intelligence (AI), but probably more significantly, much of it comes from a real improvement in computing power, and analytics tools that enable automation. I see a lot of automation of the auxiliary tasks in modelling; things like data quality checks, model performance monitoring and model documentation. There is also interest in automation of the model building process itself, but this is currently limited to sandbox environments and not something I see in production use yet.
R&C: What are the key benefits of automation in model development and deployment?
Taylor: The obvious benefit of automation is efficiency, or time savings. And efficiency has the nice quality of being quantifiable, which is important for proving a return on investment. More interestingly is the potential for improved model accuracy through building many more specific models. With automation, we can build hundreds of models in the time it used to take to build 10 models. Also interesting is the impact on decision making when models can be developed and deployed more quickly.
Apr-Jun 2020 Issue
SAS Institute