The goal of the forthcoming automation features in SAS's software is to help organisations improve efficiency and more quickly realise value from their projects.
An update to the SAS Platform announced at the company's Analytics Experience event in Milan automates data management and machine learning, as well as delivering new interpretability features.
The next release of SAS Viya, due in November, brings the latest AI and advanced analytics techniques in a form that is accessible to data scientists and business users alike, according to the company. It automates many of the steps required to transform data and build machine learning models.
Rather than hiding the process in a black box, the software produces a visual pipeline and presents the results in natural language.
Finalised models can be deployed with a single click.
The automated modelling process uses a REST API, allowing developers to customise business applications with SAS Analytics. Open source code can be easily embedded to allow analyses to be augmented with SAS.
Machine learning is applied to recommend data transformations, reducing data preparation time.
Another change is that for the first time, the November release of Viya will run on IBM Power9 systems.
Users of SAS's AI capabilities include UK insurer (and 25-year SAS customer) Admiral and Italian insuretech startup Yolo.
"We used an insurance fraud analytical engine from SAS to apply multiple techniques – including automated business rules, machine learning, artificial intelligence, text mining, database searches, anomaly detection and network-link analysis – to automatically score claims, associated entities and any corresponding social networks," explained Admiral head of business analytics Sarah Lang.
"This process has allowed us to build a strong relationship between analytics and claims fraud. A continuous feedback loop ensures we continue to update the process, identifying fraud in more cases in a faster way, whilst improving our customers’ experience."
Admiral's fraud portals eliminate the need for manual referrals, saving resources yet detecting more fraud than ever. The company has benefited to the tune of £31 million in 12 months, including £6 million in savings for claims fraud.
Yolo's approach is to use AI to create customised insurance policies almost instantly.
"As a digitally native company, our customer’s experience is paramount," said Yolo Group co-founder and CEO Gianluca De Cobelli.
"With SAS, our platform is able to process a customer’s request to insure their vacation or their smartphone using all available data in real time, allowing our financial institution and corporate partners to provide their clients with a customized and dynamic mobile experience."
SAS director of advanced analytics and artificial intelligence Saurabh Gupta said "SAS continues to innovate in AI, building on the $1 billion investment in AI we announced earlier this year."
“SAS delivers AI that helps users manage, understand and analyse the data they have and make better, faster decisions with it. The latest enhancements to our AI offerings focus on automating the many manual and complex steps required to build machine learning models," he added.
SAS is automating the analytics lifecycle, delivering strong AI capabilities, and helping transform businesses by making AI pragmatic and results-focused, Gupta told journalists during a briefing at the event.
"Today, AI is everywhere," said SAS CEO Jim Goodnight during a keynote presentation.
A growing application area is machine vision, and the presentation included a live demo of a system designed to identify the clothes people are wearing in order to help retailers decide what they should stock. That system was around 90% confident that one model was wearing Zara trousers, and around 50% sure that another was wearing a Vivienne Westwood dress.
"This is just one example of the use of new and innovative technology," said Goodnight.
Another example is the way the Amsterdam University Medical Centre is using computer vision and predictive analytics to help treat colo-rectal cancer patients. Professor of surgery and director of surgical oncology Dr Geert Kazemier explained that this is the world's third most common type of cancer, and that radiology is used to determine how patients are responding to chemotherapy – are the lesions getting smaller, remaining the same size, or growing – because it is not always possible to operate until they do shrink. The radiology process is very time consuming and subjective, but the SAS-based analytics are more accurate, saving time and lives,
"AI is very useful," Goodnight told journalists. But "there's so much hype right now" to the extent that "if you let the computer make a decision, that's called AI."
Venture capitalists are putting their money into potential rather than actual technology, he suggested. Startups are doing some useful work in the areas of machine vision and natural language processing, but the rest of AI is predictive modelling and all of that is already in SAS's products, he said.
But Goodnight said "I am looking to see a breakthrough" such as a new kind of modelling technique.
"Machines do not learn," heasserted. What happens is that we estimate the parameters of a model, try to maximise or minimise an objective function, and then keep iterating with (hopefully) improved parameter values.
"We're optimising a model, that's all... the machine isn't learning."
Goodnight also put into perspective the recent announcement mentioned by Gupta that SAS will invest US$1B over the next three years, pointing out that the R&D spend during the last two years has been about that much.
For example, the company is putting a lot of effort into software "to make it really, really easy" for people to create chatbots that work well even in situations where people use many ways of expressing the same thing.
According to IDC, SAS is once again the market share leader in the advanced and predictive analytics category, with a 27.7% market share, more than twice that of the next competitor. Additionally, SAS ranked second in AI software market share with a growth rate of 104.6%, more than three times that of the overall market.
Disclosure: The writer attended SAS Analytics Experience 2019 as a guest of the company.