What if you could see into the future? If you are running a for-profit business, wouldn’t it be great if somehow you could foresee what’s going to happen? Wouldn’t you want to make decisions that protect your business from bad outcomes and increase profitability? Wouldn’t it be great to:
-Know that your machines in the car factory will start punching holes at wrong locations at 7:21 PM when the lead engineer has gone home and it will result in a line-down event
-Know that electric submersible pumps that are pumping hydrocarbons out of your well will start malfunctioning at 04:03 AM and need to be adjusted to prevent complete failure.
-Know that the customer who just walked into your store looked at a pair of jeans and browsed through an expensive suit on your website but may be inclined to buy the jeans if you offer him a promotion while he’s in the store.
Predictive analytics empower businesses to predict what will happen next and take the right actions, just in time, to either take advantage of opportunities or avert negative consequences.
Convergence of enabling technologies for predictive analytics
Predictive analytics is increasingly being used as a viable strategy for businesses because a variety of enabling technologies have rapidly advanced and come to a point where each piece of technology can benefit from another.
Predictive analytics is based on the concept that when large amounts of data such as: machine performance, plant and production operations, data center operations, business processes, customer behavior, etc, is available, sophisticated and accurate models can be developed to predict future scenarios that have high likelihood of occurrence . These predictive models and algorithms are typically developed without any presumptions of patterns or relationships and utilize machine learning, neural networks, computations and artificial intelligence techniques. These techniques enable efficient discovery of patterns and relationships hidden in the sea of data.
In order to gain maximum value from the predictive models, businesses deploy them in real-time systems for taking actions or making decisions. These real-time systems continuously look for patterns in incoming data streams that are generated by connected Internet of Things (IoT) such as factory floor machines, assets in the oilfield, data center systems, ships in the sea and inventory pallets. The data streams are scored against rules based on smart predictive models. These intelligent systems are then able to provide answers and guidance for the toughest problems that might arise especially when a higher level of automation is needed in situations that demand fast, reliable actions that are free of human error.
Recent research by World Economic Forum (2015) reported that the most immediate value from IoT analytics and taking actions in real-time is from vastly improved operational efficiency (e.g. improved uptime, asset utilization) through predictive maintenance and remote management. Collaboration between humans and machines will result in unprecedented levels of productivity and more engaging work experiences.
A winning strategy: Predictive maintenance using TIBCO Spotfire
TIBCO Spotfire offers a complete platform for predictive analytics model development. It allows you to connect to various types of data sources simultaneously and apply advanced techniques for predictive modeling such as those that use R and advanced analytics from SAS, MATLAB and databases. Complete predictive solutions can be architected by using TIBCO StreamBase for event processing.
An example application of predictive analytics is predictive maintenance of artificial lift systems in oil and gas production. These systems are expensive and it’s highly desirable that any potential malfunction is detected before it actually occurs so that oil production is not disrupted. These systems feature real-time sensor data stream outputs for production parameters such as temperature, pressure and current. Predictive models and rules that identify leading indicators for system malfunctions are developed in Spotfire using its advanced analytics capabilities. These rules are then sent across to TIBCO StreamBase event processing platform. When the rule thresholds are violated, StreamBase sends email (or other channels) notification to the engineer with the data leading up to that violation allowing root cause analysis to be performed in Spotfire. Watch this video to learn more and check out these TIBCO Spotfire for Energy use cases.