Building Bridges

To be honest, I didn’t really know what I wanted to do at the time. It had become clear that research wasn’t my cup of tea after all, as I wanted to see fast results and impact of my work. I had been researching “data mining” techniques (yes, I like the new term “analytics” better as well) applied to chemical processes, and even though there was a lot of good stuff available on the research level, the techniques did not really gain traction in the chemical industry just yet. The cause seemed to be a combination of factors: the technology to deploy successful models in the control room apparently wasn’t mature enough and even when it was possible, engineers and operators were not really trusting the output of these (often black box) models. The analytics maturity of the chemical industry workforce just wasn’t high enough.

Coincidentally, I had a chat with Thomas and Stijn from D Square (yes, I like the new name “TrendMiner” better as well) during the last weeks of my PhD. They had noticed similar things during their data mining projects within the manufacturing industry and came up with a potential solution: giving the power of the data to the people who can interpret it. They had just started working on a new software platform that would enable engineers and operators to take advantage of the tremendous amount of data being gathered in their processes in an intuitive way, so they could start solving analytics questions themselves. That conversation really sparked my interest…

Only a couple of months later, I started a job at D Square as a Project Engineer, working on smaller data mining and alarm management projects and meeting many engineers at different companies in the process. When the first prototypes of the TrendMiner software were ready, I got the chance to start testing them with engineers at our early customers. Their feedback would have a massive impact on what became the first public TrendMiner release.

It turned out engineers didn’t even have the tools to decently slice and dice their process data and sometimes couldn’t even visualize it, let alone compare process behavior during different time periods. By developing TrendMiner as a state-of-the-art trend viewer, topped with search functionality, we gave them the tools to start making sense of their data. The introduction of the Layer Comparison Table finally led to the first big breakthroughs, solving Root Cause Analysis problems some of our customers had been breaking their head over for a very long time.

Of course, TrendMiner has evolved a lot since then, and as a Data Analytics Engineer and later a Customer Success Manager, I have been able to follow that evolution closely. But at the same time, I noticed an evolution with our customers and the users of our software. While we had to search far and wide for early adopters originally, introducing TrendMiner to engineers and operators becomes easier every day. The interest in analytics in general and self-service analytics specifically is peaking and companies are updating their analytics landscape and organization accordingly.

I strongly believe that this is a result of empowering people with the right tools. Gaining a better understanding of their process data via self-service analytics tools makes people curious about what else they can use that data for. It opens their minds for the more advanced analytics tools that have been under development for years. And while the analytics maturity of engineers and operators grows, technology must take the next step as well. I’m confident that bringing the data scientist in the loop via built-in notebooks and the possibility to deploy machine learning models directly in TrendMiner will unlock a whole new world of advanced analytics use cases.

I am happy that I got to contribute to bridging the analytics gap that seemed so large during my PhD. So, in the end my grandmother was right: engineers build bridges. Even chemical engineers do; they’re just not made of concrete.