Machine Learning And Easter Egg Hunts

Elvira Wallis

Easter is the quintessential spring holiday, full of vibrant colors, sweets, and family traditions. And yet, it may also be one of the few holidays with a built-in competition: the infamous Easter egg hunt!

It usually goes something like this: parents hide colored eggs throughout the yard and kids hunt to try and fill up their baskets before their treasures are scooped up by other seekers. It’s the only time of the year when putting all your eggs in one basket is a good thing.

As any master egg hunter knows, this is an exercise in pattern recognition and anomaly detection. You must constantly scan the landscape for anything that looks off or out of place in even the slightest way. Luckily for most egg hunters, the landscape in question is limited and pre-populated with plenty of opportunities for this activity to pay off.

But what if the landscape was exponentially much larger and the eggs were far fewer and further between? You may be just as interested in those Easter eggs, even if your childhood egg hunting strategy no longer works as well.

So it goes with interpreting data that comes from the Internet of Things. While all the sensor data that comes from the Internet of Things can give your business a new level of detail into day-to-day operations, it doesn’t necessarily follow that those hidden Easter eggs (read: insights) will present themselves automatically.

Enter machine learning and its ability to perform tasks by learning from data. For example, imagine for a moment that you are an operations manager for a chocolate company that makes those chocolate bunnies you might have consumed this past Easter holiday.

Creating chocolate bunnies requires lots of perishable products to be stored and available in an optimal condition for the standard chocolate bunny process to be successfully completed. The issue is that once a problem is detected (say, a deviation from the specific viscosity required for molding the chocolate into chocolate bunnies), it may already be too late, with a batch ruined.

However, details surrounding that detection could allow the system to learn from this experience. It could be that the viscosity issue has a higher probability after other smaller anomalies (such as temperature fluctuation) have occurred in succession. Anomaly detection could indicate when a larger failure might occur.

And anomaly detection could be further explored through influencer discovery. Meaning, once the system determines a correlation between smaller temperature anomalies and a subsequent viscosity change, the next step would be to discover what led to this pattern.

This is where a bunch of statistical analysis comes into play. Essentially, because the Internet of Things allows for so many data points to be collected over time, a digital system can correlate different combinations of those data points, which can include details on various environmental variables.

Imagine that a viscosity deviation that might ruin a batch of your favorite chocolate bunnies may be a function of outside temperature, humidity, capacity utilization, pressure, power usage, flow, etc. If each of those indicators is being tracked and measured over time, machine learning can take all that data, identify patterns across it, and provide much more actionable guidance on the influencers discovered.

Enterprise business processes can be extended through connecting the previously unconnected goods or products. This treasure trove of data available through connected goods enables you to quickly react to changing conditions or resolve issues the moment they arise, not only through real-time visibility into product inventory, state, and utilization, but also through having the underlying data directly integrated into existing business systems.

Now, combine that visibility and integration with machine learning capabilities, and you get additional options for optimizing the enterprise. This enables you to maximize the value of your products while also mitigating any future negative influencers that could impact your day-to-day business operations.

This type of business outcome is precisely what connected goods with embedded machine learning enable. Product insights are derived from key indicators that are continually strengthened by machine learning. A digital system with machine learning can help you detect conditional anomalies and discover influencers for abnormal product quality deterioration. Simply put, think of it as the ultimate Easter egg finder for your business.

Want to learn more about how to maximize product value in a connected world through IoT-enabled inventory, storage, utilization, and consumption insights? Visit our connected goods web page.


Elvira Wallis

About Elvira Wallis

Elvira Wallis is the Senior Vice President and Global Head of the Internet of Things (IoT) at SAP. She is responsible for ideating, defining, delivering, and taking to market IoT business solutions to increase revenue, adoption, and thought leadership.