agsandrew - Fotolia
A recent study by Capgemini highlighted a gap in the implementation of operational big data – but some utility companies now have the infrastructure to make the most of operational analytics.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
The study found that although 54% of UK respondents routinely collect unstructured data – the highest proportion of all countries surveyed – and 44% of UK organisations use external data to enhance insight, the UK showed the lowest rate for utilisation of operations data.
Mark Deighton, principal, insights and data, at Capgemini UK, said: “The UK has done a lot of work on building BI infrastructure, but only one-third of UK organisations have made analytics a key success of their business.”
Data warehouses are well established in business, providing a qualitative approach to analysing structured data. “They apply quality control at the point of ingestion to address known situations and are aimed at producing management reports,” said Deighton.
But for operational analytics, data needs to be ingested from disparate manufacturing systems and open data feeds. “You don’t have control,” he said.
Rather than deploy a data warehouse, companies use data lakes to bring everything in, said Deighton, adding: “You only attempt to structure data when it is time to query.”
Among the sectors that benefit from operational analytics is utilities. “We are looking at using telemetry data from water pressure valves and meters to predict maintenance across the water mains network,” said Deighton said.
By incorporating social media and GIS data, the water company can identify whether there is likely to be a leak in a particular area, he added.
British Gas has been fleshing out its operational analytics programme two years and Daljit Rehal, strategic systems director at Centrica, said the utility company went live with its first application about a year ago.
Read more about operational analytics
- Few manufacturers have successfully mined their operational data to improve their business processes.
- Big data analytics processes on data from sensors and log files can propel users to competitive advantages, but a lot of refining is required first.
The company deployed a Hadoop cluster to create a data lake to pull in data from its SAP billing and customer care systems. “British Gas is an energy business and a homecare insurance business,” said Rehal. “We have a federated approach for analytics as different parts of the business live on some form of analytics.”
British Gas began by creating a data processing framework, said Rehal. “We needed to bring in data from several hundred systems into one place, ready to be consumed.”
To achieve this, it used Hortonworks Data Platform (HDP), a distribution of Hadoop as a data lake.
Previously, it needed six hours to extract data from 100 SAP tables to load into its data warehouse, but only had a four-hour nightly batch window, which meant the data was not up to date.
The Hadoop cluster enabled the company to extract data without having to wait for SAP to go down, said Rehal. “We are always up to date now and can ingest data from 20,000 SAP tables from SAP, available in our Hadoop cluster,” he added.
Having this data readily available means central heating engineers can tap into CRM or billing data to get the latest information before visiting a customer’s home. For example, said Rehal: “If our engineers have an iPhone, we have an app that gives them the latest information on the customer.”
British Gas is also working on real-time decision-making and proactive fault-finding to enable engineers to understand the customer better. For instance, the engineer can see whether the customer was on hold on the phone for a long time, said Rehal.
Link traditional IT with operations
Data science is also being used to link traditional IT with operations. Rather than treat an engineer’s laptop as an IT asset, said Rehal, it can be considered as a remote monitoring device, such as using its 4G modem to check availability of mobile data.
Such insight can be fed into the smart meter systems as part of a root-cause analysis if meter readings at a customer’s home suddenly stop being collected, he said.
Operational analytics enables British Gas to predict demand, said Rehal. “We have a model to understand how many boilers will break down, so we can have the right number of engineers ready,” he said.
In February, Mark Hodges, Centrica’s chief executive, energy supply and services, UK & Ireland, discussed the benefits of operational analytics as part of the company’s Hive connected home product.
“It’s really this integrating of the devices, the ‘hero’ products and the hub through the data analytics and trying to figure out a way to monetise that in terms of a revenue stream, which we see as the real prize,” he said.
Capgemini’s report Going Big: Why Organisations Need to Focus on Operations Analytics said: “Operational analytics can make organisations more productive and smarter. But achieving this potential rests on a number of key principles: a robust data strategy focused on making the maximum use of data and making analytics an essential part of the decision-making process in operation.”
But although many organisations have built data-collecting infrastructure, they have yet to gain the benefits from analytics. One Computer Weekly reader noted recently: “Businesses need to invest in better decision technology. All the good data is worth nothing if it doesn’t result in better decisions being taken.”
Since the benefits can extend across many parts of an organisation, as the British Gas example shows, operational analytics needs to be driven from the top down.