Subscribe

Data without analytics is a wasted asset


Johannesburg, 03 Mar 2021
Donald Farmer, principal of Treehive Strategy.
Donald Farmer, principal of Treehive Strategy.

The biggest mistake that businesses make when it comes to using their data is not using it. Businesses do not think of data as an asset that they can glean real insight from, and will help them to make better decisions.

ITWeb Business Intelligence Summit 2021

Register now for ITWeb’s annual BI, data, analytics and AI event. It is an invaluable platform to engage online with senior decision-makers, and hear from local and international experts on how to embrace digital transformation to create a data, analytics and AI driven culture. For more information, and to register, click here.

“They think of it as a form of record keeping, which is a big mistake. All businesses today are data businesses. From the smallest corner shop in Africa, to the largest enterprise, if the businesses keeps accounts and books, they have data.”

So says Donald Farmer, principal at Treehive Strategy, who will be presenting on ‘Transforming data into future insights using predictive analytics’, at ITWeb Business Intelligence Summit 2021, to be held from 9 to 11 March as a virtual event.

“If you have data, and you're not analysing it, then your data is a wasted asset, and just there as a means of record keeping. If you analyse it, then you start to take advantage of it, and you can take action. Data businesses need to be analytics businesses, and if you're an analytics business, you need to take action based on those analytics."

Looking into the future

Data analysis, says Farmer, is based on existing data within the organisation. It is always based on the past, things that have already happened. “We have records of customer sales, insurance claims and suchlike. It's always, in a sense, backward looking.”

However, businesses want to know about the future, and predictive analytics enabled them to take that backward-looking data and project it into the future and make certain assumptions and guesses about what will happen, he adds.

For example, this enables businesses to plan better, explains Farmer. “If you're a retail business, you might be able to plan stock levels more accurately using predictive analytics. If you want to make the best offers to new customers, then you can use your knowledge of existing customers, project that into the future using predictive analytics, and suggest what new customers would like.”

And so, in a sense, Farmer says predictive analytics enables businesses to answer questions about what will happen in the future. “I think for all of us as human beings, we're curious about the future, we’re planning for the future, and are worried about the future. Predictive analytics helps us to address these issues to a certain extent, and not only technical and business issues, but the very human issue of anxiety, concern, planning and worrying about the future. And excitement about the future too, for that matter.”

Don’t aim for perfection

Speaking of the pain points to avoid when implementing predictive analytics, Farmer says the biggest pain point is always trying to make the model as perfect as possible.

“I always think of it this way: if I can tell you with 100% certainty that it's going to rain, it’s probably pretty obvious because everybody could tell you that. In some ways, a model which is too perfect, which we call overfitting the model, is actually not very useful. This is either because it's too obvious what's going to happen, and anyone can do that, or because the model has been manipulated too much.”

Avoid trying to make your model perfect, he advises. It should be good enough to give business an advantage. Take future sales, for example. Future sales are not predicted down to the rand, but down to the thousands, or tens of thousands of rands. It doesn't need to be 100% precise.

“Sometimes people aim for too much precision, rather than for practicality, so a compromise needs to be reached.”

As human beings, we're not very comfortable with ambiguity, adds Farmer. “We like clarity. We like precision and accuracy, and we’re uncomfortable with ambiguous statements that may not be absolutely precise. We need to get more comfortable with that. So the pain point is really our human discomfort with dealing with ambiguity, and data scientists and people who work in predictive analytics get around that problem by analysing the ambiguity and being very technical about it. However, a level of comfort needs to be reached.”

Skills, attitudes

During his talk, Farmer will give delegates an idea of the skills and organisational attitudes that people need to develop in order to effectively use predictive analytics. “This is the team that you should be hiring, the skills you should be looking for, and the skills you should be developing internally. Also, this is how your organisation should think about predictive analytics and the way in which it can be used.

“Predictive analytics isn't as difficult as you might think. It isn't some dark science that requires hugely technical capabilities. In fact, as business people, we make predictions all the time. It’s not that difficult to get started,” he ends.



Share