Hard though it is to say (and to believe), I joined SAS 20 years ago this week, on Tuesday 14 April 1998. As a fresh-faced, bright-eyed 23 year old, I worked in the UK training department as a statistics trainer. SAS 6.12 was the latest version of our product, SAS/Assist was a revolutionary way of getting a system to write code for you, and SAS Enterprise Miner was in beta version, ready for its first release in 1999.
Back then, I had no idea that I would still be employed by SAS in 2018. I had a tendency to get bored quickly, and thought I would move on within a very few years. SAS, however—or perhaps more properly, analytics—has provided me with all the change and interest that I could ask for. And there have indeed been a large number of changes over the last 20 years! Analytics has very definitely become ‘cool’, with the advent of artificial Intelligence, data science, and machine learning, but what other changes have we seen?
Data structures have become a lot more varied
Twenty years ago, most of the data used for analysis was structured, and was either from well-defined experiments or transactional data from operational systems. Now, we do a lot more with text and speech, and image analysis has moved from the academic world into business.
Awareness of analytics is much higher
Artificial intelligence is now mainstream, judging by the number of times the phrase is mentioned on BBC news. My friends and family have become more interested in what I do, with a few even choosing to work in this field, though I don’t want evangelise too much about analytics as a career, because we all know what happens when demand outstrips supply.
Expectations of analytics are massive
As awareness has been raised, so too have expectations that analytics will deliver huge value for organisations, with new programmes popping up all over the place. More creative ideas are being used in applications, because the impossible is now becoming possible. It is great that analytics is seen as a core business capability, but there is also a danger that people come to expect “magic” from both individuals and technology. This may lead to disappointment and disillusion before AI has reached maturity.
The terminology has evolved, and continues to do so
One of the biggest changes in the last 20 years is in terminology. In 1998, we manipulated data so that we could analyse it statistically. We stored data as subjects with corresponding variables, in tables with rows and columns, using it to try to find relationships that would improve decision-making. Now we are wrangling data to allow us to build artificial intelligence solutions, with machines that learn automatically. Like any language, the dialect of analytics changes over time, and I often feel that one of my key roles is to translate between business and analysts as we work together to solve business problems.
The world now operates much more quickly
The general pace of the world and life has dramatically increased. There is pressure to respond instantly to texts, emails, instant messages, and tweets, or risk offending someone. My daughters expect certain levels of speed, functionality and value from their technology, which I certainly didn't at their age, and they are not unique. Individuals and organisations expect answers and solutions quicker than ever before, and analysts have to respond to that.
A new generation of experts is emerging
When I left university, your typical statistician or analyst had a background in mathematics or statistics, with a strong grounding of theoretical knowledge. We used technology in our work, but we weren't necessarily experts in programming. The current crop of analysts have much more diverse backgrounds, with less theoretical knowledge, but a stronger emphasis on their technical skills, so they can scale up data science routines across a range of programming languages.
Many more vendors now offer analytics capability
This is probably both a blessing and a curse. It is fantastic to see a strong marketplace with open source vendors, new entrants, and traditional vendors all offering different ways of analysing data. This has meant the pace of change has been rapid and there are many great solutions. However, it is also creating confusion for consumers, who struggle to cut through the hype, and find the right solution. The world of analytics feels very different from 20 years ago, but still as exciting and challenging as ever. In 20 years’ time, I expect to be looking back from a completely different world again—and also hope that I will still feel as excited about my work as I do now.