How a Cycle of Data Analytics Enables Operational Excellence

May 02, 2019

There are many innovative technologies being leveraged by EHS and operational teams to successfully anticipate and mitigate risks. One of these is data analytics.

Do a search on “data analytics” and you’ll end up with millions of results (about 70-75 million on Google depending on the day, if you’re searching on the exact term). “Analytics” has become another frequently-used buzzword.

This post will explain the different types of analytics and show how greater value is achieved when they work together.

Four Main Types of Analytics

Let’s start by giving some definitions of the different types of analytics. There are many models out there, but I really like the analytics framework from LNS Research (you can see a visual representation of it in this article). There are four main types of analytics (actually there’s a fifth type, cognitive analytics, but we’ll focus only on four in this post):

Descriptive Analytics: This type of analytics looks at the past. It analyzes and aggregates data from past performance and events to answer the question “What happened?”.

Diagnostic Analytics: This type is somewhat an extension of the previous one. Some models simply combine the two, but many keep them separate. This type analyzes data to answer the question “Why it happened?”.

Predictive Analytics: Whereas the two previous types of analytics focus on the past, this one focuses on the future. It uses historical data, forecasting models, statistical probabilities and other techniques to answer the question “What will happen?”.

Prescriptive Analytics: Just as diagnostic analytics comes after descriptive analytics, we can also say that prescriptive analytics comes after predictive analytics. It uses simulations and data on how past, similar events were successfully handled to answer the question “What should be done?”.

Whenever you read or hear about “big data analytics” or “advanced analytics”, ask yourself which type of analytics is being discussed. Similarly, sometimes someone may erroneously say “predictive analytics” when they may be in fact talking about one of the other types of analytics.

Bring All Types of Analytics Together

Each of the different types of data analytics brings great value on its own. But there’s an even greater value when all four are used together as part of a cycle, which leads to operational excellence.

Start with descriptive analytics to look at past incidents and identify the areas of your organization that are experiencing the most adverse events. The areas could be either specific plants, processes or occupations. Take into account both the frequency and severity of adverse events. For example, a plant experiencing three severe incidents a year may require more attention than one experiencing six very minor events a year.

Follow up with diagnostic analytics to dig deeper and understand what are the most frequent causes for the most recurring adverse events. This allows you to develop corrective action plans to eliminate hazards or implement controls to reduce risks emanating from exposure to these hazards.

Looking at past events provides valuable insights. But sometimes past performance may not be indicative of future performance.

There may be plants or processes that have not yet experienced major incidents, but that may be prone to one because of warning signs, or because they have similar characteristics to other plants or processes that had problems in the past. This is where predictive analytics is useful. It will help you identify potential high-risk areas so you can predict and prevent incidents.

After you identified high-risk areas that may have problems lurking just below the surface, you can leverage prescriptive analytics to help determine applicable actions plans or control measures to address the risks.

Finally, go back to the beginning and descriptive analytics again to analyze adverse events and detect patterns, at a regular frequency.

The key to enable a cycle of data analytics that leads to operational excellence is to have a software platform that:

  1. Captures and aggregates large amounts of machine-generated data through sensors, beacons and drones connected via the IoT, in addition to data entered by humans through desktop and mobile interfaces. With Industry 4.0, there will be greater communication of physical assets/systems with each other and with humans. Anticipate a greater share of machine-generated data.
  2. Offers advanced technical capabilities such as artificial intelligence and machine learning.
  3. Generates valuable insights from data to proactively predict and prevent incidents.

When looking at analytics, be sure that you’re considering all types, that they work together in a cycle, and that you have the right technology.




Jean-Grégoire Manoukian

Content Thought Leader