polygraphus - Fotolia

Automated predictive analytics on the rise, says SAP’s UK chief technologist

Mark Darbyshire, SAP UK and Ireland’s chief technologist, reflects on the supplier’s updated predictive analytics software and service pack 10 for Hana

SAP UK and Ireland’s chief technologist Mark Darbyshire said the supplier’s recent announcements “profoundly underline how Hana is a data platform, not just a database”. 

“Our predictive, real-time analytics means a CIO can do something for the business today in a way that either users or business operations can consume.”

At an SAP Insider conference in Nice last week, the supplier announced the general availability of the 2.2 release of its predictive analytics software and a service pack – 10 (SPS10) – for its Hana in-memory database platform.

Darbyshire said of the most recent iteration of the predictive analytics product: “Customers can get predictive insights on ultra-wide datasets, where there are tens of thousands of columns, or you have complicated or unstructured datasets.”

He said the update will support data scientists better and make the process of extending their algorithms to a broader business population at a customer organisation more automatic.

“Expect to see more profound support for automated machine learning,” he said, and confirmed that SAP has UK customers using analytics for predictive maintenance on machinery and for “next best action” suggestions to increase sales.

SAP turns a corner

New capabilities in the 2.2 version of “SAP Predictive Analytics” include those for internet of things (IoT) applications, improved integration with Hana and enhanced integration with the open-source statistical analysis language R.

The service pack for Hana has significance for the data platform “playing better in a Hadoop environment and in IoT applications", said Darbyshire.

“Hadoop clusters can become stores of static one-off brilliant insights in the data laboratory. But how is Hadoop being leveraged by the rest of the organisation? Hana is a nice way for the rest of the business being able to take advantage of Hadoop. You can send data from Hana to your Hadoop cluster or the other way round for analytics.”

Read more about automated predictive analytics

  • How companies are hitting a wall in the use of predictive analytics, with lack of time and resource cited as their biggest problem.
  • Manufacturers are realising that long-term sustainability requires the ability to exploit machine-learning algorithms by employing predictive analytics.
  • Consult this guide to call centre analytics.

Enhancements for Hana are said to include faster data transfer with Spark SQL and a single-user interface for Hana and Hadoop cluster administration using Apache Ambari. The supplier also announced "expanded smart data integration capabilities of Hana to the latest Hadoop distributions from Cloudera and Hortonworks".

IT industry analyst Holger Müller commented on his blog: "A year ago mentioning Hadoop in SAP circles was a bad word. It is remarkable to see SAP turning the corner here and supporting the most prominent enabling database technology for next generation applications. Spark was already mentioned at Sapphire and will be key for SAP to make the future offering fly.

"It is also key for SAP to support both Cloudera and Hortonworks on the distribution side, SAP cannot play favorites here – and should not either. 

"And it's always good to see SAP use common, open-source tools like Ambari".

IoT at enterprise scale

Last week, SAP said the SPS10 for Hana will deliver new capabilities to get customers connected with the IoT at enterprise scale. A remote data synchronisation feature in Hana brings in data from the edge of a customer organisation’s network. Developers will be able to build IoT and data-intensive mobile apps using SAP’s SQL Anywhere suite, said the supplier.

And so, for example, data can be fed back from retail outlets and restaurants to headquarters, and sensor data on ships, pump stations and in mines.

Darbyshire said: “We are doing that now in the real world, as opposed to talking about it. 

"This represents an evolution. How do you support the range of IoT device scenarios? Some are only intermittently connected, or truly remote. 

"The simplistic view of IoT – where everything is always on and connected – is not always optimal from a resource point of view. 

"We’ve also introduced better spatial data support, and better data mining and text analysis.”

The supplier said it now has 6,400 customers for Hana, double the figure from a year ago. The Hana cloud platform has 1,400, it said. And S/4Hana, its full ERP on the platform, signed up 370 customers in 2015.

Read more on Big data analytics