How to use machine learning to drive your business

0

Machine learning (ML) has become a buzzword during the last few years, attracting business leaders across various sectors to reinvent their processes with the help of artificial intelligence (AI). Making decisions based on what ML solutions have learned has become a prerequisite for running an innovative business. Brigitte Naylor-Aumayer caught up with Pasi Helenius, Principal Data Scientist at SAS, to better understand what business leaders need to know about ML to drive business growth.

Machine learning is all the rage. Are there businesses or industries that could particularly benefit from it?

It will touch every industry, but I’d like to think about those organizations or sectors that are up against data native startups and cutting-edge companies that are quickly rising to become significant competitors. Operating in such a competitive market forces businesses to modernize their operations. Investment in ML-powered solutions is a step they take to build their competitive advantage.

Many of those traditional organizations are, however, struggling with ML. What makes it so challenging?

ML is complex technology. Looking under its bonnet, one can find a massive variety of methods and algorithms that might seem confusing.

Could you shed some light on how business leaders can understand and use ML to deal with disruptive competition and drive business growth?

The first step really does not have much to do with understanding the technology. I recommend at first to take a closer look at business needs. Knowing the options innovative technology might deliver can be inspiring. If you keep an open mind, it could lead you to drive innovation yourself or to try something entirely new in your business area.

So how do you get started? Looking at use cases?

If a technology looks exciting, you might start looking for potential use cases. Ask yourself if any of those could work at certain parts of your organization. However, moving in this direction – from the algorithm to the business – is tricky. Keep in mind that technology should match and amplify your business processes and not shape the core of your business operations. In my experience, the best first step is looking at your core business with an open mind. Assess all of the decision-making processes in various areas across the organizations, from manufacturing and supply chain to customer interaction and marketing. Then consider how your teams could use insights from data analytics and machine learning to support these decision-making processes.

Keep in mind that technology should match and amplify your business processes and not shape the core of your business operations.

Some decision makers might find that this comes with a lot of organizational and probably also cultural change. What if they cannot show reasonable benefits for their business straight away? Should they still invest in machine learning?

If you don’t see a clear benefit at the moment, think about what will happen in five to 10 years when a data native competitor or startup will make those decisions with the help of machine learning and achieve better results. Just imagine someone entering your sector with a data native culture, looking at every decision as an analytical, data-driven problem. If you don’t prepare for this beforehand, you might be in trouble later on.

Let`s move on and assume the decision for ML has been made. You have mentioned yourself that the debates about how to build an ML system sound confusing. Is it really that complex?

Consider the matter from the perspective of a data scientist. For a data scientist, all algorithms are just tools in a toolbox. When building a solution, you pick the ones that suit your business needs and the available data. Then you apply them to generate insights or explain and interpret your results. When building an ML system, it’s equally important to choose the right algorithms and ensure appropriate data flows when putting the model in operation.

To improve your decision-making processes, rethink the entire journey. Consider your processes against those of a competitor that is data native. It’s not enough to optimize one part of your business or build an individual AI application. Sure, such a solution will increase your bottom line, but analytical decisions need to be made whenever possible throughout the business. All of your decisions are analytical problems – and your processes should reflect this.

Still, the path from ML research to applying ML in production environments is riddled with various threats. Any advice on how to resolve it?

Having a solid foundation for business applications of machine learning will determine your success. Implementation of ML is challenging even for organizations that boast a mature engineering strength. To set your business up for success, you need to ensure that it has an infrastructure that enables robust research and supports production applications. Cloud technologies are an essential part of this effort because they enable teams to deploy ML models to high-volume production environments. A solid infrastructure also helps in large-scale testing of models and frameworks, allowing you to expand the experimentation with ML across the entire organization.

So let`s assume we have assessed our business processes with data in mind, as well as all the decisions made on the way and the infrastructure our ML applications would require. What changes should we expect for our organization?

To be able to run your business in the best way, your decisions need to be based on data and algorithms. This is what will enable you to compete with data native startups that are sooner or later going to disrupt your sector. To accomplish that, you need an organizational structure to support this and – above all – a culture that embraces analytical, fact-based decision making. So treat machine learning as a toolbox and pick the tools you need to make your business thrive, thanks to better decision making.

AI and Machine Learning are key technologies for building a competitive advantage in sectors marked by strong competition from data native startups. Find out how to navigate your AI journey by visiting the AI Pathfinder.

 

Share

About Author

Brigitte Naylor-Aumayer

I work as a content and communications professional in EMEA region helping customers find marketing insights with analytics.

Leave A Reply

Back to Top