In a recent article from MIT Sloan Management Review, Jeffry D. Camm and Thomas H. Davenport examined how enterprise machine learning and predictive analytics programs have been disrupted by the coronavirus pandemic. As they put it, “The data we use to make good managerial decisions has been caught up and turned upside down in this unpredictable marketplace.” Companies are struggling to adapt and redeploy their models and strategies for the new economic environment, but how should they move forward with this process?
The authors offer a number of steps for rebooting data science initiatives, among them:
Weigh Data Relevancy
Organizations must determine what to delete, what to keep, and what to impute. Many of the experts with whom Camm and Davenport spoke suggested using moving averages and other smoothing forecasting techniques to determine how much to rely on pre- and post-pandemic data.
Embrace Increased Use of External Data
According to the article, “Trying to model low-probability, highly disruptive events will require an increase in the amount of external data used to better account for how the world is changing. The right external data could provide an earlier warning signal than what can be provided by internal data.”
Ramp Up Model Auditing and Stress Testing
Keeping a close eye on machine learning and prescriptive analytics models is an essential part of post-pandemic data science initiatives. As Camm and Davenport put it, “Techniques developed for quality control in industrial engineering, like control limits and acceptance sampling, need to be applied to machine learning to make sure the models are ‘in control.’”
Construct a Portfolio of Specialized Models
If there is one business lesson emerging from the coronavirus, it’s that things can change at a moment’s notice and the importance of flexibility and agility cannot be overstated. When it comes to data science, companies should “consider developing scenario planning and simulations to construct specialized models that can be ‘pulled off the shelf’ as needed.” A second lockdown remains a strong possibility, so analyzing what the organization learned since the initial outbreak and determining how these lessons could be applied should there be a second wave is a critical step in strengthening the overall data science program.
As the article puts it, “It’s a different world from the relatively stable data and analytics world of the past.” In this environment, organizations must find a way to retool their prescriptive analytics and machine learning initiatives so that they can make the best decisions.