Machine Learning: Smarter Data Analytics Improve Decisions
When it comes to technology, the only thing we can be certain of is that things will change — and they will probably change rapidly. For example, just think about how different things were before the Internet exploded onto the scene about 20 years ago. Future historians will probably include the Internet alongside such inventions as the printing press, the cotton gin and the automobile as technologies that radically changed the world at that time.
There are several new technologies today that have the potential to soon be world-changing. MIT Sloan Management Review recently published an Executive Guide1 that focuses on seven technologies that are changing the world. One of these that really caught my attention was machine learning and augmented, automated data analysis.
Components of Machine Learning
The Executive Guide identified three main technologies that are critical components of machine learning:
- Cloud computing — This has resulted in the separation of data storage and processing from devices themselves, which has created ubiquitous access to data and software while also allowing new ways for people to collaborate.
- Big data — The generation and collection of large amounts of data (both structured and unstructured) is allowing organizations to uncover new insights and responses to challenges they face.
- Artificial intelligence (or AI) — AI programming and algorithms enable digital devices to access and share data so users can explain and forecast events, trends and processes.
When combined, the impact of these three technologies goes well beyond the realm of information technology. They form the foundation of machine learning by making data in all different forms more readily available. Users can now access data from practically anywhere in the world, using a plethora of device types. And the data can now be effectively analyzed and applied.
Of course, the ready availability of so much data can be a two-edged sword. Data can come from both inside and outside of an organization and be both structured and unstructured (think tweets, social media posts and blogs). While the sometimes-chaotic flow of data can be inconvenient at best and paralyzing at worst, it does present a tremendous opportunity for organizations to gain new insights not only into their own operations, but also those of their customers, suppliers and other key stakeholders.
Moving from Non-Interpretive to Interpretive Decision-Making
The Executive Guide emphasizes the importance of extending your explanatory reach from non-interpretive (or highly predictable) decision-making to interpretive (or unpredictable) decision-making. This is the best way to realize the full potential of machine learning as a new way for decision-makers to address complex chains of cause and effect, notes the Executive Guide. The guide goes on to explain four types of analytics1:
- System analytics are for monitoring and control of semi-predictable processes and systems.
- Predictive analytics are used as input and context for decisions with high uncertainty.
- Control analytics are for monitoring and control of predictive processes and systems.
- Process analytics are used as input for decisions with semi-high uncertainty.
If viewed on a grid, system and predictive analytics together would be considered strategic analytics while control and process analytics together would be considered operational analytics. The key to effectively putting machine learning to work, according to the Executive Guide, is expanding your analytical range from operational analytics to strategic analytics. Or in other words, shifting your mindset from one of monitoring and control to one of forecasting and planning.
If your organization is anchored in control analytics, you could be overly exposed to competitive risks — particularly if your industry is highly competitive and volatile. By harnessing the power of machine learning, your organization can expand your capabilities to include system, process and predictive analytics to “gain access to deeper insights and accelerate your decision cycles,” notes the Executive Guide.1
The core outcome of machine learning technology, the guide concludes, is “learning in all its myriad forms.” The evolution of data availability, analysis and access technologies is causing a shift in many of the assumptions that underlie learning theory. Machine learning tools like simulations and virtual reality give organizations a more complete understanding of complex systems while allowing leaders to shape content and outcomes to meet diverse needs.
Among the new technologies that have the potential to be world-changing soon are machine learning and augmented, automated data analysis. Cloud computing, big data and artificial intelligence form the foundation of machine learning by making data in all different forms more readily available. Users can now access data from practically anywhere in the world, using a plethora of different kinds of devices. As a part-time CFO or project CFO, a CFO partner from a CFO services firm can help you determine both how your organization could benefit from machine learning and how to implement an effective initiative.
1 Seven Technologies Remaking the World; by: Albert H. Segars; MITSloan Management Review; March 9, 2018
Arthur F. Rothberg, Managing Director, CFO Edge, LLC