When you read the words ‘machine learning’ what is the first thing that comes to mind? You’re not alone if your answer was Artificial Intelligence (AI). Machine learning and AI has been a staple of science fiction for decades and it is perhaps this link that makes it so difficult for us to picture a world where it is a possibility.

The thought of intelligent machines that can learn for themselves is a little dystopian. Digitisation is common the supply chain, but surely we will never reach a point where decisions are made without human input?

The problem is that technology never goes backwards and neither do customer expectations. As soon as something can be done more efficiently, it will be, and machine learning potentially represents a huge step forward in making processes better. The more this technology develops the more machine learning stops being a possibility and starts being a probability. There are even early examples of it being successfully implemented in some companies. We now need to look past science fiction and dig deeper into what machine learning really means.

What is it?

Machine learning is a way of analysing big data that gives computers the ability to learn without being explicitly programmed. Current big data algorithms calculate outcomes based on data sets, but this always requires some human input. This might be adding new data sets or tweaking the algorithm. The core concept of machine learning is to give machines all the data and the tools they need to tweak themselves. In theory, a person would just need to set this up and then leave it to work.

Is this the same as AI?

Not quite. Artificial intelligence is the broader concept of machines being able to carry out tasks that we would consider “intelligent”. Machine learning is about giving machines data and using AI to let them learn for themselves.

Why is this beneficial?

In many cases, it is not, and human intuition still trumps machine learning. But there are a few examples where people have struggled to keep up with the pace of change.

Demand forecasting is one example. Market patterns fluctuate quickly and reacting to this change can mean the difference between maintaining a good market position or making a costly error. Machine learning algorithms can be modelled dynamically to react to the market. They can even take into account other big data sources like social media which human-driven algorithms might not include.

This should not be seen as an alternative to people. Machine learning is a just method of producing more accurate data to give people the tools they need make the right decisions.

Where has it been adopted successfully?

There are a few companies that have used machine learning, but it should be noted that it is in the very early stages of adoption in the supply chain.

Grananrolo, an Italian dairy company used machine learning to increase their forecasting accuracy by five percent and decrease delivery times by half. Groupe Danone (based in France) also used machine learning to improve the forecasts of their promotional offers which were in the past shown to be up to 70 percent inaccurate.

Should I be investing in it?

Not yet, it is still too early. Gartner predicts that mainstream adoption of machine learning is at least five to ten years away. But it is worth starting to collect the kind of data you would need to make it work. As we have discussed in previous posts, big data analytics can have a big impact on your supply chain. Machine learning is a different way of analysing big data, but it will be much easier to implement if you already have the data sets in place.

Machine learning at the moment is a far cry from what we have been exposed to in science fiction and this is perhaps for the best. Machine learning is not designed to replace people but give them the best possible tools to make the right decisions.

Want to know more about us? Get in touch by filling in your details at the bottom of this page!