I recently spoke with an executive who has a $200 million problem. His company supplies materials to a manufacturing firm that keeps a tight lid on its inventory. This lack of visibility is forcing my customer to create excess material, without knowing exactly how much the customer really needs. I asked this executive, what if I could save him 30%, returning upwards of $80 million in working capital to his business? (read on)
Welcome to the age of predictive analytics, where data-driven insights can predict outcomes today, tomorrow, and into the future.
How? Data science.
Data science is where “scientific methods, processes, and systems” collide with data-producing technology to extract insights. In an industrial context, that means capturing data streams from connected assets so you can track, measure, and predict machine behaviors.
Take a jet engine, for example. Sensors can detect when there is an increase in vibration of turbine blades. With predictive analytics, you can plan scheduled maintenance, which can greatly reduce unplanned downtime across the fleet of aircraft, minimizing risk. You can also predictively ensure that spare parts are precisely where they need to be! And that’s just one of many illustrations of how predictive analytics can move your operations from reactive to proactive.
The Tipping Point
How, you might be asking, can we truly predict outcomes with data? Two key developments have helped spur the advance of predictive analytics.
The first is a rapid increase in qualified data scientists. In the last five to seven years we’ve seen an influx of graduates with a mash-up of expertise in statistics and computers science entering the workforce. These one-part seers and one-part scientists began to create tools that could crunch and understand huge volumes—terabytes—of data, providing the technical capability to predict and solve the problems of the future.
The second development was deep learning—the ability to efficiently train computers to pinpoint what to look for, and what not to look for. For example, we can train computers to read a radiology image to look for an area of interest on a scan (e.g. a tumor). The computers learn to kick back findings, freeing the radiologist and physicians to do more valuable things—such as diagnosing and treating disease.
Reducing Risk: Predictive Data in Action
GE Digital is in the middle of several exciting projects in industry. One of them is with a major university medical center on the East Coast of the United States.
We approached this project with the customer in three steps.
First we used historical data to looked back on their operation. Using analytics, we looked at data to identify average wait times in the waiting room and trauma center; the number of patients treated, etc.
Second, we then used data to show them insights into today. The data reflected what was happening right now. For example, the teams got to look at triaging patients in real-time, assessing key questions such as, does the facility have enough beds, and if not, what’s the nearest facility that could accommodate those patients, and so on.
Finally, we gave them a view into the future. With predictive analytics, we could show them what would happen in next 12 to 18 hours. The insights into the past and the present gave them a higher level of confidence when looking at the data about the future.
Despite major leaps in data science through the combined power of analytics and the cloud for compute processing of massive amounts of data, there are still holdouts. From my experience, companies dealing with very sensitive data or highly regulated industries are often least receptive to using the cloud. The pharmaceutical industry, for example, guards its IP very closely, and tends to prefer to contain it within their four walls versus using the cloud. So, we work to educate these companies in cyber security, on encryption, and how we could use algorithms to predict the success or failure of a clinical trial, using cloud technologies as the engine to crunch massive volumes of data.
It’s important to examine how the growth of advanced analytics will affect jobs. As the radiologist example shows, jobs may change, but that does not mean they necessarily go away. We see this across all industries, but it does require that organizations adapt.
Where should you begin using predictive analytics and machine learning to reduce risk?
Determine what can be delegated. Some tasks have a simple answer with clear delineation—yes or no, black or white. Let machines handle these rote tasks, and save skilled resources for more difficult, complex actions.
Do you have data, and if so, do you practice data liquidity? In other words, can that data be moved to better perform needed analysis?
Investing in the future
A Forbes Insights report reveals that over the next two years, more than half of global executives’ plan to invest at least $10 million in advanced analytics resources. It makes sense. Having the right data means you’re able to make the right decisions, such as reducing unplanned downtime, which significantly reduces risk to your business. And it frees up your skilled talent to spend time on larger issues facing your business.
At GE Digital, we’re working closely with customers to demonstrate how these new capabilities can solve industrial problems and increase the performance of assets. We think this growing field of data science will lead to the broad expansion of predictive analytics across the industrial sector over the next few years.
Oh, and as for the executive with a $200 million problem on his hands? He's eager for any help we can provide to return working capital to the business.
For more on the subject, check out a report from Forbes Insights and EY. You can also check a perspective from GE and our partner EY on this report here.