Prescriptive Analytics – Actual Business Intelligence

By Adam West, first year Pearson Business School student

Adam is currently undertaking an internship at Satalia. Satalia is a thought leading prescriptive analytics company that uses the latest optimisation algorithms from academia to solve the hardest problems in industry. During his internship Adam has written a great blog on ‘Prescriptive Analytics – Actual Business Intelligence’, so we thought you may like to have a read!...

Think about the last time you went to the doctors… they’ll have taken a look at your medical history, your current symptoms and made a prediction about what will happen next. Doctors describe the past, predict the future and prescribe treatment to best deal with what the future holds. Albeit a far removed, humanised example – this is the current situation across industries.

Imagine if your doctor described your symptoms and predicted their effect – but then passed on the prescriptive decision making process to the less qualified, less knowledgeable receptionist. It wouldn’t quite work, and you’d be left feeling dissatisfied with the whole process.

Organisations employ tools that describe and predict but not those which allow decision makers to take data-driven actions. A recent PwC study of 1,100 senior executives found that ‘data and analytics’ is the third most important factor when making big decisions; behind their own intuition and the experience of others. This makes sense considering current ‘data and analytics’ do not actually help to solve problems; they simply help describe them and forecast the possibility of them happening in the future.

Most Organisations

  • Gather Data (Big Data – structured and unstructured)
  • Analyse data and describe what has happened, what is happening now and why (descriptive analytics)

Fewer organisations

  • Are using a tool or software to identify insights, trends and patterns in data and create predictions of what will happen next – including some nice visualisations (predictive analytics) Gartner estimates that only 30% of industry use some form of predictive analytics tools
  • Use either an house built or commercial algorithm that solves the problem to a certain extent

Very few organisations

  • Make truly data-driven decisions
  • Use machine learning, artificial intelligence or adaptable algorithmic tools that provide actionable recommendations from the insights provided (prescriptive analytics) Gartner predict a 3% adoption of prescriptive analytics

This level of adoption comes as no surprise, Gartner’s “hype cycle” puts prescriptive analytics at the peak of inflated expectation; early adopters are experiencing a mixture of success stories and disheartening failures but the majority of organisations choose not to adopt. See the image below and the following link for more details about Gartner’s methodologies.

Adam West's Blog

Prescriptive analytics will eventually change the way business are ran. Here’s what needs to change for this to happen.

Advancements in AI and Machine learning

Many employees do not yet fully trust machines to make critical business decisions. Employees entrust machines to represent the data in a way that aids the gut decision of a business decision maker, but not much further. It’s a reasonable precaution, business is contextual; a world which combines technology, maths, behavioral psychology and uncertainty – a game which humans believe they can win and they can, but not without help.

Advancements in machine learning will vastly improve the machine’s ability to learn and adapt to the vast multitudes of data they receive. Meanwhile, the maturity of the ‘algorithm economy’ will see huge competition (and improvements) in algorithm providers that help solve business problems. The sooner these advancements occur, the sooner prescriptive analytics will become a trusted and integral industry standard.

Man and (not vs.) Machine

A recent study by the University of Queensland identified that humans ability to process and understand variables within the decision making process is limited to four variables, with five-way problems being performed no better than chance. Despite this, the most complex and significant part of the analytical process; “What shall we do now?” is largely performed by humans, only small parts of the big decisions are understood and actioned upon.

Naturally, decision makers will be stubborn to relinquish power and the fear of redundancy is a huge challenge for all ‘semi-fully’ automated technologies – especially when it competes directly with senior decision makers. Current decision makers (that tend to let emerging data scientists deal with the maths and technologies of current analytics) will become decision scientists. A term coined for those who possess expertise in technology, maths, business decisions and behavioral science. These newly defined, expertise hybrids will facilitate the decision processes and take action in the face of uncertainty (see decision theory) with a clear data-centric bias from the prescriptive tools they help build.

For prescriptive analytics to become the norm, organisations will also have to acquire and adopt: bigger datasets, systems that process structured and unstructured data, synergistic predictive and prescriptive analytics, algorithms that adapt and leaders that trust. When an organisation is able to harness all of the above effectively, they’ll find they have an edge over their competitors – and prescriptive analytics will be the latest trend that comes to fruition. Only then, will organisations achieve true business intelligence.

Interested in studying Business from within a FTSE 100 Company? Check out the Pearson Business School section of our website.