I am fascinated with how technology impacts the way we live and work. Innovations like the Internet and the apps built on it have affected nearly every important part of our lives. The world has grown smaller and everything has become increasingly connected.
These big societal changes are transforming industries as well. For example, the manufacturing industry increasingly automated otherwise manual processes during the 20th century. This automation has made companies faster and more efficient at producing their goods.
We’re now almost two decades into the 21st century and we’re on the brink of another big shift in manufacturing. As the Internet allowed everything to become more connected, companies started gathering the data of their machines. But processing and analysing vast amounts of raw data requires significant time and effort from humans. Algorithms can do this faster and more accurately.
Humans can understand the correlation between two points if there are a few obvious variables. A second-hand car costs less because of its mileage, its age, and the fact that it’s been used before. But when there are more than a few explanatory variables, or when the relationship between those variables isn’t immediately obvious, it becomes much more difficult for us to assess why there is a correlation between two points.
It’s hard for humans to accurately predict when a supermarket will run out of whole milk, because there are simply too many variables to take into account. But an algorithm can crunch much more data, and can take into account many more variables to find correlations that humans cannot.
Additionally, while humans are experts at creative thinking, if a situation can be loosely defined in a broad framework of rules, algorithms are much faster. This is why an airbag saves lives on a daily basis. During a crash, the crash sensor of the vehicle sends information to the airbag electric controller unit at such a speed that the airbag can be deployed in less than 1/20th of a second, much faster than any human could ever react to the impact of a crash.
But despite the significant added value that algorithms bring, and despite some gradual adoption, I still see some reluctance towards algorithms in my daily job. This isn’t unusual. After all, implementing an algorithm means giving away control to something that’s often seen as a black box. Although an algorithm might produce results quickly and accurately, it feels uncomfortable that we often don’t understand how it came to these results.
Additionally, algorithms cannot be introduced with a snap of the fingers. They need to be carefully integrated into existing processes, which doesn’t just mean that it needs to be installed and connected properly, but also that it needs to be accepted by the people that will work with it.
I believe that the key to successfully implementing algorithms is by starting small and scaling up fast once you’ve concretely experienced the benefits. Take a tiny step. For example, implement a detection algorithm that notifies an engineer when a bearing needs to be replaced.
This will have the benefit of reducing the time engineers need to spend in diagnosing equipment. Additionally, the algorithm doesn’t take any action. All it does is notify the engineer, who can then still check whether the algorithm was correct or not.
Once you trust this detection algorithm, you can expand it to notify the engineer of different faults on the same machine and of faults on different machines too. The tiny benefit of an algorithm that detects whether a bearing is wearing out will now become the big benefit of an algorithm that can detect multiple faults in different machines.
I think that we’re not far from a tipping point where companies will have no choice but to participate in this new trend of not just gathering data, but optimising and acting on it as well. Other companies are implementing algorithms and if you don’t, you will fall behind. Although this might sound like unpleasant news, it also means the opposite: Algorithms can offer tangible and near-immediate benefits to your company.
So my advice is: Start small, but don’t remain small. Scale quickly once you see the benefits, otherwise you risk getting stuck with several small solutions that operate at suboptimal efficiency. A centralised algorithm that can find correlations between many different areas will help propel your company to new heights in the 21st century.
-Morris Beek, Technical Consultant at Widget Brain
Labour standards play a crucial part in making sure a business knows how many minutes of work to schedule based on an expected demand. It is an important pillar of the roster next to having an accurate forecast, but what if you don't have labour standards? Learn more about the good and bad approaches to define labour standards.
Each one of your stores, restaurants and/or locations has its own context: different demand drivers, local events and seasonal effects. And they all need their own forecasting method to give you the most accurate results. Hyperlocal forecasting allows you to do that. Read more about the why, how and what of hyperlocal forecasting.