Business intelligence as a whole is moving from a “sense and respond“ model to a “predict and act” model. The following section explores analytics from an asset management frame of reference.
Gartner has defined an analytics continuum, demonstrating how increasingly mature analytics offers a greater competitive advantage. As illustrated below the analytics continuum is split into seven milestones. The milestones progress from raw data collection to implementing optimisation processes.
To embark on the path illustrated below businesses require cleansed data to drive various reporting and analytical systems and processes. That is not to say businesses require a perfect data infrastructure to begin producing valuable insights and results. Today’s tools can handle a variety of data sources scattered along the maturity path. A comprehensive data strategy and its accompanying infrastructure is the end goal, but businesses will miss countless opportunities while cost justifying the perfect system.
Maturity Milestones (Gartner)
The four phases of analytics are described below:
1. Descriptive analytics: This type of analytics looks at historic performance. Common examples of descriptive analytics include Key Performance Metrics (KPIs) such as availability, failure reports etc. Valuable reporting succeeds in combining data from disparate sources such as financial and operational reporting.
2. Diagnostic analytics: Diagnostic analytics endeavours to interpret data and extract valuable insights. While descriptive analytics simply reports on e.g. equipment failures, diagnostic analytics looks for common drivers among failures by overlaying additional information such as environmental and operational characteristics. Powerful diagnostic analytics has become accessible in recent years with the introduction of tools such as MS Power BI and Tableau. These tools offer a wealth of functionality with a manageable learning curve. The objective of diagnostic reports often informs what data the business should be collecting and how.
3. Predictive: Predictive analytics represents the logical progression of diagnostic analytics. If implemented properly, predictive analytics allow business to answer the why. For example, why did the piece of equipment fail, or why does this type of bearing last longer? If a business understands the drivers of performance, they can anticipate. Predictive analytics incorporates statistical-, machine learning(ML) and mathematical models that attempt to predict what is likely to happen based on the known information. Predictive analytics is a rapidly expanding field, but it is not yet as accessible as quality diagnostic analytics.
4. Prescriptive analytics: Predictive analytics allows business to anticipate, informing an appropriate response. In comparison, prescriptive analytics goes one step further by recommending appropriate responses based on what is likely to happen. In order words, prescriptive analytics is a decision-making tool. It considers what is known (available input data, business rules and constraints)and suggests a best suited response to resolve what is likely to occur with or without intervention.
For example, a prescriptive system can be used to model the failure of electrical transmission poles and recommend appropriate maintenance activities. With a predictive system It’s usually up to the subject matter expert to determine whether it’s worthwhile replacing pre-emptively or retrospectively based on its location and consequence of failure. A prescriptive system shortcuts the decision-making process by considering business rules and costing information and recommending a response conscious of all the trade-off. In other words, a prescriptive system attempts to recommend an optimal or optimised response based on codified SME decision making logic.
Predictive maintenance is maintenance that bases the decisions around what is performed and when, on some prediction about future states of an asset.
This terminology is commonly used in industry, but is a little misleading. The consensus is that predictive maintenance is maintenance that is informed by a predictive system and analytics process, but has little to do with the maintenance itself, and more to do with the overall strategy or approach for addressing one or more failure modes. The maintenance itself is not predictive in any way, e.g. greasing a bearing at a certain timing could be informed by a predictive analytical process, or by an Original Equipment Manufacturer (OEM)recommendation. The maintenance task (execution) does not change, although the justification for these timings may be different.
The key elements are shown below.
Predictive analytics in a predictive system informs maintenance
The predictive system can be anything from conceptual, to paper based, but is typically a computerised system running some sort of calculation software. The predictive system executes a method or process to predict a future state, and this is called predictive analytics. The additional information generated from this prediction informs better decisions around maintenance task selection and timings.
Asset -"a useful or valuable thing or person"
Predictive-"relating to or having the effect of predicting an event or result."
Analytics- "the systematic computational analysis of data or statistics"
When combined, our definition of asset analytics is:
"The systematic computational analysis of data or statistics relating to an asset and its future operation and performance."
In it's simplest form, predictive analytics uses known information to make a future prediction about an asset. The prediction could be related to the number of failures, the expected condition or even performance metrics.
Predictive analytics has two parts;
A predictive system is any system which can facilitate a predictive analytics process.
The terms "asset analytics" and "predictive analytics" are very broad, and take many forms. An analysis has a specific objective, and it's own set of requirements and constraints. The purpose of performing an analysis is usually to answer one or more questions about the asset in its business context.
Predictive analytics for the basis for future decision-making, based on data and current understanding.
Some examples of these questions could be:
Each of the above questions can be answered by one or more types of analysis, however we believe that they should ALL be analyzed using the same working knowledge base (See knowledge management here)
Each type of analysis can be performed using any number of methods to reach the desired answers. Some approaches utilize basic statistics, machine learning, big data, Artificial Intelligence, Causal Diagrams, or any other method currently available to achieve its purpose or answer specific questions.
Predictive analytics provides a way for organisations to anticipate changes, and make better strategic decisions.
Asset predictive analytics is about making predictions. These predictions are data and/or assumption based, and forecast future events. These forecasts can then be used to make informed decisions around an organisations operations, and subsequent performance.
Some use cases are:
The ability to see what is coming and make strategic adjustments, is what separates the reactive businesses of today, from the industry leaders tomorrow.
Predictive analytics is also a foundational element for prescriptive systems, which can recommend courses of action under different scenarios and constraints. Once a method for prediction is established, variances can be introduced to answer the "what if" type business questions.
Criticisms of predictive analytics
There are three common points of criticism when it comes to predictive analytics.
Our list of solvers and calculators is continually growing.
 K.Lepenioti, A.Bousdekis, D. Apostolou and G. Mentzas, "Prescriptive analytics: Literature review and research challenges", International Journal of Information Management, vol. 50, pp. 57-70, 2020.
 A.Goyal et al., "Asset health management using predictive and prescriptive analytics for the electric power grid", IBM Journal of Research and Development, vol. 60, no. 1, pp. 4:1-4:14, 2016. A
"A Brief History of Business Intelligence - DATAVERSITY", DATAVERSITY,2021.[Online].
Finlay, Predictive analytics, data mining and big data : myths, misconceptions and methods. Steven Finlay. Basingstoke, Hampshire: Palgrave Macmillan, 2014.
 Evans, James R. &Lindner, Carl H. (March 2012)."Business Analytics: The Next Frontier for Decision Sciences".
Want to find out more? Contact us here.