Whether monitoring automated equipment, instigating finance solutions or constructing workflow patterns, predictive analytics enable businesses to plan ahead as best they can. Dr. Sheela Siddappa, chief data scientist and global head – data analytics at Bosch, passionately believes businesses large and small can do much more than they currently are with predictive analytics, arguing that while analyzing the data can help fix a faulty piece of machinery, it is always better to look at the bigger picture and adopt a "prevention is better than the cure" approach.
Ahead of her presentation at the Predictive Analytics Innovation Summit in Chicago, Innovation Enterprise sat do with Dr. Siddappa to discuss the utilization of predictive analytics and AI tools, the talent gap within her specialism and how businesses can improve efficiency through an effective predictive analytics strategy.
Innovation Enterprise: Can you provide our readers with an insight into your presentation at the upcoming Predictive Analytics Innovation Summit in Chicago?
Dr. Sheela Siddappa: While there are organizations that have already implemented predictive analytics strategies, I'll be exploring the bonanza of tasks that can be carried out, above and beyond what people currently think predictive analytics is capable of.
For example, for predictive analytics in manufacturing, the common belief is that the maximum we can stretch to is predictive maintenance, with the main use of predictive analytics in manufacturing focusing on a component of interest and maintaining it.
Whether it's focused on vibration sensors or temperature sensors, we can collect all of this data, move into the cloud, analyze it and then create a visualization from which we can plot decision-making mechanisms.
If you look at vibration patterns on a piece of machinery and it deviates slightly, you will be able to deduce if that pattern will lead the component to stop working, which we can then act upon earlier.
If I follow the norm, that act can be asking an engineer to fix the problem. But if I look at the big picture, then there is a possibility of giving that engineer much more time to prevent a problem, and prevention is better than the cure, which is what I'll be covering during my presentation.
IE: What predictive data trends do you expect to dominate the industry in 2019?
SS: Today, organization are starting to understand the power of predictive analytics and are collecting data from selected machines, locally, in that machinery plant. I think we'll see a scaling up of the amount of data we can collect from machines, as well as adapting standard techniques to allow us to mimic scenarios across plants located across different geographies. I see that as being a key trend in 2019.
Today, I don't see it happening – it is more localized to a given plant, and every plant is trying to analyze how they can improve efficiencies from predictive analytics. However, I see this scaling up from 2019 and think we'll see some localized success stories, which the centralized heads will want to embrace and implement in other plants, which is why I see adoption being really high over the next year.
IE: What is your wider view on the adoption of predictive analytics throughout the business community?
SS: At Bosch, we have embraced predictive analytics and have predictive maintenance in place. From a data scientist or machine learning perspective, we have begun to understand this technology; we've trained our models with our own industries and customers and we are already learning. Generally, from a global perspective, this is a trend in manufacturing.
However, while there are organizations that have already understood the benefits of predictive analytics, who are ready to scale it up through investment and implementation, there are many other organizations that really need to start thinking about predictive analytics on a much bigger scale. Many can see that predictive analytics is the next big thing, so naturally the adoption rate of predictive analytics solutions will grow exponentially.
The push back, however, tends to come from the fact that it can represent some investment. Depending on the types of technologies businesses want to adapt or monitor, it can still be simple, with a brief impact on performance – a position from which they can choose to scale up adoption.
Visit Innovation Enterprise's Predictive Analytics Innovation Summit in Chicago on October 30–31, 2018
IE: What risks do you see arising from the growing use of predictive analytics – particularly in respect to AI – when it comes to making important business decisions?
SS: One potential concern is practical information that is not practically available to the algorithm, which means decisions being made are not reflecting all practical situations. Depending on the domain and the decision the AI is taking, it could be misleading because it is not the complete information a human would pick up through their sensory organs. For example, machine production level information, performance, inventory, schedule and demand information may be available, but information related to an upcoming water or power shortage in the geography which is very essential for production, or an unexpected change in weather condition, are not considered in model building today. However, I would not worry about the future of AI as I think it's going to be really good. From a talent perspective, it's going to be a fascinating place in which to work because the AI will focus on the operational efficiencies beyond the mundane, repetitive tasks, while humans can focus on new business strategies, disrupting some of the exciting areas and innovating concepts every day.
IE: What advice would you offer a large organization that is in the process of setting up new predictive analytics team?
SS: Today, many organizations have a head of analytics who does not have a fundamental background in analytics. If you're looking to develop a solid analytics strategy, then the head of analytics should have a background in analytics, or at the very least have worked for some time in the field, as they need to be able to understand what is feasible and what to expect.
Once a great leader has been identified, I think everything else will more or less will fall into place.
IE: How big is the talent gap within the predictive analytics field and what issues is that gap creating?
SS: There is a big gap. The pressures are that while there is a good level of students and graduates coming into the industry, we often want to employ someone with a bit more experience and who is mature enough to understand predictive analytics from a business perspective. From an academic perspective, solutions can often be quite substantial and take a lot of hard work and time to implement. However, from a business perspective, we want things to be simple. We don't take pride in saying that we've implemented the most complex algorithm; we take pride in making things simple and practical when solving a business problem – from a usage (comfort and convenience) and long-term point of view it works much better than a complex solution.
The moment people enter the industry, there are very few people well-placed to guide them, nor is there a manual they can follow because historically these were not worked on. This is new to industry and each domain and use case has to be independently thought through, analyzed and then provided with the solution. We need talent that has common sense and logical thinking over and above the fundamental theoretical skills.
Siddappa Sheela, head of global delivery – data analytics at Bosch, will be speaking at Innovation Enterprise's Predictive Analytics Innovation Summit in Chicago on October 30–31, 2018.