New world, old model. What’s Next?

By Soumitra Dey ‘Sonny’
– Data Consulting Partner, Ampersand Advisory

to the Movement Control Order(MCO) we have seen an incremental consumption of
non-perishable items like toilet papers / hand sanitisers, leading to a spike
in sales and profits for stores. But once the world returns to normal, this
trend may not continue, thereby complicating the data.

Present Day
In the wake of extreme and unforeseen changes brought on by the global COVID-19 pandemic the world is changing around us, and the deployed machine learning models are losing their power, as current behaviours in the economy shift and change rapidly in these uncertain times, organisations and their data teams are rushing to update their models.

Supply Demand out of Sync
One of the key concerns among many is that their current models could be generating inaccurate or misleading predictions. Because machine learning models are based on historic data to predict future events, the COVID-19 pandemic presents a unique challenge for data scientists because there are no similar events for comparison that can inform competent predictions.

Understand the Changes
A starting point is to take a big step back and ask which actions the models are impacting and how operations could change as a result. Travel, retail and oil prices, for example, have all changed significantly in the past few weeks alone, so it is critical for organizations to assess which features are most crucial to their models. Once these changes are in place, organizations can obtain a more realistic understanding of where their new business priorities should be, as they contend with constantly changing circumstances.

Predicting missed
appointments for medical outpatient visits isn’t as important right now. On the
other hand, building models to allocate bed/ventilators for patients is
currently far more vital.

what features the models should be considering is essential to making the
necessary changes and adjustments. These insights can help organisations
pinpoint which operations they should prioritize.

Model Sensing

than monitoring models it is important to monitor changes to the models’ input
distributions, to consider how data drift could be impacting findings, and
understand how these changes affect their overall accuracy; identifying areas
of fragility and aggressively changing them to become antifragile. These
insights from new input data can help data science teams and business leaders
pinpoint which operations they should prioritize. Doing so will accelerate
several technological trends such as:

  • Marketing & Finance functions will find that their existing data are no longer relevant. Only when new data is mashed with existing ones will this lead to better scoring models as well as information about current search terms, social media behavior, and even changing demographics, which will provide a more accurate picture of what today’s shoppers care about, and who they should prioritize.
  • Corporate risk functions leveraging on external data that can be mitigated with corrective actionable insights and patterns and avoid the pitfalls that could severely impact their bottom line in uncertain times. 
  • With the rise of cloud and algorithmic intelligence, digital representations of a supply chain no longer need to be a patchwork of models for sourcing, manufacturing, distribution, etc. Instead ‘Digital Twin’ a living model of a supply chain can be digitally rendered and real-world events can be simulated to predict outcomes and actions for decisioning.


honed during the 2008 crisis are facing the ultimate leadership test in the current
crisis. Unlike natural disasters, viral epidemics unfold over time, are unique
in their epidemiological characteristics and have a far higher degree of
uncertainty with respect to the overall cost in lives, productivity, economic
output and business revenue.

industry has been immune to the impact of this event. For some, like grocers,
disinfectant manufacturers and food delivery services, the temporary sales
spike data trends are beneficial, not injurious. Other industries, like
airlines, hotel and convention operators and video conferencing services, have
seen such profound disruption, both positive and negative, that it could result
in permanent transformations.

depends on how rapidly business leaders can filter the transient noise from the
persistent trends since there are indications that some industriewon’t return
to their former states. Given the breadth and magnitude of social, personal and
business disruption, there is a higher than normal probability of permanent
changes as many areas of society and business resets to new norms.

business landscape will be littered with companies that couldn’t see or adjust
to permanent, systemic shifts in business processes, consumer behaviour, and
societal norms. Organizations that correctly ascertain and capitalize on their
data and AI model trends will have a better chance to return to their pre-crisis

Usher in the new Model

events like the novel Coronavirus crisis are outside our wildest expectations
and ability to statistically model, they are inherently unpredictable,
destabilizing and psychologically traumatizing. Business and organizational
leaders must resist the panic impulse while simultaneously juggling two
critical tasks:

  1. Quickly adapting to short-term disruptions to minimize financial and personal disaster doing what’s required to live to fight another day.
  1. Assessing the long-term cultural and catalytic changes that affect its products and services to position the organization to thrive under new circumstances.

leadership requires not reverting to what might be familiar or comfortable, but
sticking to long-term business priorities that are relevant in a post-crisis
world. It also means embracing changes like new AI led data models that provides
a competitive advantage under new business rules.

situation offers leaders a clear path to begin navigating to the next normal

It’s time to look outside of our own silos for the answers and it’s time to make sure that despite the crisis we face, we can give our organizations the power to make it through them. The question now is, will your organization adapt to find its way out of the unpredictability and how?  Using the tools that can give you better data, as well as the means to predict and chart a course through the uncertain landscape is no longer optional, it’s mandatory if you want to thrive and not just survive.