Guest Author: Lee Sacco, Senior Director, Applications Development at Oracle Corporation
While many businesses have gotten good at using predictive analytics to make predictions, not so many businesses have figured out how to get measurable benefits from those predictions. Predictive analytics makes predictions about what is likely to happen, while prescriptive analytics tells us what we should do about it. Competitive advantage comes from learning how to turn predictive analytics insight into prescriptive analytics action.
How does a business go from insight to action? Prescriptive analytics can take a decision point in a process, use predictive analytics to determine the most likely outcome for each decision option, and recommend actions based on which has the best likely outcome. Prescriptive recommendations guide employees to actions in line with the goals and values of the business, and ensure that important insights aren’t accidentally missed or intentionally ignored. Prescriptive is an especially useful tool when the business goals are well defined and the prescribed tasks for each decision option are finite.
Thus, prescriptive analytics is a tool very well suited to a repetitive, process-intensive business domain like asset maintenance. Asset maintenance has well defined goals; for example, most maintenance operations need to maximize availability, minimize failures, and optimize costs. Asset maintenance has well defined roles, workflows, decision points and action plans. And every day, many times a day, specific decisions need made for each asset.
- As a maintenance planner, I need to decide when to maintain each asset, what tasks need done and which parts need replaced at each maintenance interval, so that I can meet reliability targets at optimal cost.
- As a maintenance manager, I need to decide if there is a risk of warranty fraud or non-compliance with safety, data or environmental rules, so that I can reduce fraud claims, improve worker safety, and avoid non-compliance penalties.
- As a parts planner, I need to decide how many of each spare part are needed in which locations and when, so that I can maximize first-time fix rate and reduce spare parts acquisition and holding costs.
- As a maintenance technician, I need to determine root cause of failures, decide on the best fix, and determine whether an asset should be repaired or replaced, so that I can minimize turn time, reduce repair cost and eliminate re-work.
The same decisions need made for each asset, but each asset’s unique configuration, history, usage, environment, conditions and parameters determine the best decision option and action plan for that asset. Prescriptive analytics ensures that these decisions are made taking all available data and insights into consideration, that the right person has the recommendations and insights needed to make the right decision, and that the information is incorporated into the workflow seamlessly right when the decision needs made.
The Enterprise Apps Advantage for Prescriptive Maintenance
The standard predictive analytics process: spend a lot of time figuring out what business problem to solve, what data you need to solve the problem, how to prepare the data, and how to make use of the predictions. And spend most of the project preparing the data.
But enterprise apps have specific advantages for predictive and prescriptive analytics. Enterprise apps solve specific business problems. Enterprise apps capture the data needed to solve those business problems, which is generally also the best data for making predictions. Enterprise apps standardize business processes into role-specific workflows with predetermined decision points and action options, making it simple to deploy predictive insights and prescribed actions to the exact spot in the workflow where they are most valuable. In short: enterprise apps provide an ideal prescriptive analytics framework.
A perfect example: the Oracle Maintenance Cloud enterprise app. Maintenance Cloud was built to solve problems like optimizing maintenance costs and minimizing asset failures, problems well suited for prescriptive analytics. Maintenance Cloud captures maintenance-related data needed for predicting asset lifespans, failure rates and maintenance costs, including technician notes, chats, log files, meters, sensor readings and counters. And the Maintenance Cloud application has built-in tools that transform data to be readily combinable with IoT data streams and readily digestible by machine learning algorithms. Predictive analytics together with the enterprise apps framework enable prescriptive maintenance.
Below are a few examples of maintenance-related predictions and how those can support prescriptive recommendations in Oracle Maintenance Cloud:
Predict that an asset is about to fail
Recommend auto-creating a work order and dispatching a technician
Predict reliable lifespan of a part
Recommend replacing a part before it reaches end of lifespan
Predict remaining useful life of an asset
Recommend retiring rather than repairing old assets when they break
Predict lifetime maintenance cost
Recommend a maintenance plan with optimal lifetime maintenance cost
Predict root cause of failure for an asset
Recommend best fix to repair asset based on root cause
Predict parts demand at each work area
Recommend part stocking levels at each work area
Predict work order turn time
Recommend escalating a work order to avoid missing the due date
Predict repair cost
Recommend replacing the asset rather than repairing it
Predict warranty fraud
Recommend opening a case to determine if claim is fraudulent
While the predictions seem magical, it is the prescribed actions that minimize unplanned failures and optimize maintenance costs; and turning predictions into profitable actions is where the real magic is.
To learn more about predictive analytics for prescriptive maintenance, check out the Cloud Customer Connect webcast “Predictive Maintenance Isn’t Magic…or Is It?”.