In my last post, I talked to Prenton Chetty, Head of Analytics at Nedbank, about the impact of analytics on the business, and how modelling has changed the conversation and improved decision making. This post continues our discussion, and explores Prenton’s views on how data science is changing with AI, and some of the potential applications.
From a personal perspective, what sparked your interest in AI?
More and more, the business was asking us to solve problems that just could not be touched using traditional methods. There were situations where we didn’t have the right tools, so we needed something new. I think for us that was the biggest driver: We just needed to try something else, and AI and advanced analytics seemed the obvious options. We piloted a few and saw the increase in accuracy, and then it became a no-brainer.
Tell me about open source, how you’re using it, and the value that you’re getting out of it.
Open source has its place, definitely, and I do see huge advantages, but I’m also a big SAS® user. SAS has now integrated open source, which is brilliant, so we can still use Jupyter Notebook and integrate it back into SAS. I think this is excellent. SAS cannot be beaten on ease of use, speed to deliver, and things like that. It allows you to build extremely fast and be very accurate. With SAS® Viya®, we now also have the visualisation package. This is a big piece for us because now analysts can get the data, build the model, get some graphs out, put that into a presentation and go straight back to the business with a solution. Open source gives us a lot more ability to fine-tune certain things, though. There are likely to be bespoke situations where open source will be essential, and it also gives us the option to try techniques that SAS hasn’t fully integrated yet.
Beyond the traditional customer analytical models, what new type of problems are you solving, and how are you looking to incorporate AI?
We always try to push the boundaries and find new techniques, and new ways of using old techniques. So, for example, we’re trying to translate voice calls to text, but accuracy is not great. In South Africa, with bad lines and connections, and the various accents and languages, it’s quite a task, but we’ll keep going. We are already using text analytics, especially around complaints data, to identify our customers’ pain points and how can we resolve those. Image recognition is another thing we’re playing around with, but not in the traditional sense of facial recognition. Instead, we’re trying to create an image of the client's transactions by translating values into a colour or a density of colour, and turn each customer profile into an image. We can then use image recognition to see whether we can identify unique characteristics of this client, and if we can show how the image evolves over time, and then predict what will happen next. It’s very early days, but it’s really interesting stuff.Beyond the traditional customer analytical models, how are you looking to incorporate AI? #ai #DataScience Click To Tweet
AI, deep learning and machine learning are very hot topics at the moment. Where do you see AI having the biggest impact outside banking?
I think it’s going to have enormous impacts. Any area where you need some sort of planning, predictive analytics, or to understand a scenario, it can be used. My wife is a doctor, and I can think of a million ways it could help her and her colleagues. People are already writing image recognition stuff that’s reading X-rays with insane accuracy, and it’s also being used for cancer detection. On the engineering side, IoT is making it easier to predict a machine failure, and so helping with planning and maintenance. You could even use it to predict the stock market, and there are plenty of people trying. You don’t need 98 percent accuracy – even 70 percent accuracy will make money. There will always be people trying to figure out how they can predict something better and get a competitive edge. It looks to me like very much a first mover type of scenario.