Delivering Business Value using Machine Learning
Updated: Dec 15, 2019
“Subject matter experts (SME’s) are well placed to identify low-hanging fruit that are ripe for picking with intelligent, autonomous systems.”
The global race is underway to become a leader or ‘fast-follower’ in digital, autonomous, intelligent systems. Large companies across technology, energy, manufacturing, healthcare and consumer goods are already seeing tremendous value in establishing digital teams and functions within their organisations and working with external providers of artificial intelligence (AI) and machine learning (ML) software. Digital products and tools can be integrated into almost every function within an organisation to identify opportunities to improve safety, efficiency and reliability. The opportunity to improve business operations with AI and ML is sparking creative solutions within organisations and it’s enabling the revitalisation of previously uneconomic projects or endeavours.
One of the keys to successfully integrating digital solutions into the day-to-day lives of employees is being able to link the business need, problem or opportunity, to a digital solution. In order to make this link, you need to collaborate. When software engineers, data scientists and subject matter experts (SME’s) combine their knowledge, experience, intuition and creativity, significant value can be realised.
What happens if data scientists don’t ask the right questions when developing a system or software?
Developing autonomous and intelligent systems requires an integrated, collaborative and multidisciplinary approach. The success of a digital project or initiative hinges on effective stakeholder engagement.Subject matter experts (SME’s) are well placed to identify low-hanging fruit that are ripe for picking with intelligent, autonomous systems and can help with prioritising different digital projects. A data scientist may have created a robust and scalable model to be distributed to a business function, but if even one constraint or unrealistic assumption is made, the whole project could be viewed as invalid.
What happens if SME’s don’t ask the right questions when developing or using a system or software?
It’s important that when SME’s are presented with a new digital product, they understand how the algorithm was developed and what data was included when training, validating and testing that model. Of course, the user-experience of the tool is also important and that is often the focus from the people who sell or develop the tool. Important questions to ask your digital solutions colleague or provider could include:
Which SME's have been involved in this project?
What type of machine learning method did you choose?
What classical or deep machine learning models did you consider?
What sensitivities were run to determine the most robust model?
What metrics were used to determine the most fit-for-purpose model?
What data was used? (Then ask yourself, is that representative of the entire sample space?)
How was the data acquired and ‘cleaned’ for use?
What machine learning methods did you consider?
At SIG, we're transparent in how these algorithms are developed and how the algorithms are integrated with data to create specific models. We encourage all clients who develop their own ML solutions or engage external providers to feel comfortable asking these questions. After all, the underlying assumptions and ML technique can make all the difference!
The Artificial Intelligence winter has been and gone and know it’s time to watch as advanced computing capabilities and software are able to be easily integrated into existing business models. The AI spring is here, where value is blossoming in major industries globally. Machine learning, the internet of things (IoT), edge computing and cloud computing are being integrated into business models and it’s exciting to think about where we’ll be in 5 years as we continue to push the digital frontier.
Have you got any questions about machine learning? Connect with us to find out more.