The COVID-19 crisis has meant that a lot of businesses are walking in uncharted territory. This disruption has profoundly changed the way many businesses are run. Almost every client of Intellify has been affected by this once-in-a-generation change to the way we do business in Australia.
The MIT Technology Review noted that the COVID-19 crisis has…
“Affected artificial intelligence, causing hiccups for the algorithms that run behind the scenes in inventory management, fraud detection, marketing, and more. Machine-learning models trained on normal human behavior are now finding that normal has changed, and some are no longer working as they should”.
Intellify has been helping our clients leverage machine learning technology for Forecasting (inventory, sales, real estate, insurance claims) for some time now, and we’ve found the techniques have great advantages. The highly non-linear methods available through Gradient Boosting (GBM’s) or Neural Networks means that forecasts can be highly – and tightly – tuned to the historical data.
However, few of today’s companies existed during the last major pandemic in the 1920s, so there is little actual historical data on how pandemics affect modern business processes (sales, inventory management, etc.) – which means that even the most sophisticated of these models still have limited ability to model the changes which have hit us all this year.
But there is a second, more subtle impact on the models from the COVID crisis. Even once the step-change has hit a business, and its forecasts, there is a lingering effect – the future forecasts will also be dependent on historical data – and that historical data will now include the data affected by the COVID-19 crisis!
Intellify modeling with our clients shows that the lingering effects on future forecasts can adversely affect model accuracy for many years in the future, although of course, the exact effects will be different for every model and every client.
Our experienced consultants at Intellify have put together a toolkit of techniques that we can use to diagnose problems and identify possible ways to rectify them. Then we can use a range of cross-validation techniques to show how the rectifications impact the modified model forecasts.
We’ve seen first-hand how this change has played out and we know the time to adapt is now. Contact Intellify at email@example.com today to find out how we can help your organisation.