The next generation of analytic models is reshaping big-data projects, as we know it today - for better
I remember the time when learning about using a computer pertained to coding a piece of software, which resulted in the loss of understanding for a majority of the audience. Computers were so complicated to learn to use, that associating them to geeks was a notion that prevailed for a long time. However, when ‘normal’ people started using their PCs to produce - instead of coding - the world saw a spike that didn’t exist before. Today we can make the same parallel with analytics. And this is exactly where Augmented Analytics comes in, changing the game – for better.
Analytics is about solving business problems, in almost real-time, by utilizing the insights – along with data - that the operations and clients generate constantly. It tells the full story of a business, without having to go through the tedious records one by one.
To begin with, the main question is what exactly ‘analytics’ is? To elaborate, analytics is a process of data analysis and interpretation done either by humans or computers. While the task might be tedious or impossible - due to the amount of data - for the former, the latter does it more efficiently and in a much more precise way, making it almost mandatory for companies to invest in the field – in fact, 97% of the U.S. based companies are investing in it already.
Why is it so important? To quote the statistics, about 90% of the world's data was generated in the last two years. The data is so extensive that it would take approximately 180 million years to download all of it from the internet web, and that’s just the beginning. Can you imagine how much data our clients will have in the next 10 years? This makes it evident that the use of more advanced analytics is the ultimate solution for all types of businesses to prosper.
As it evolves, Analytics expands to different tasks, typically classified based on their function, including descriptive, diagnostic, predictive, and prescriptive. In other words, a look at the data can help extrapolate and predict future scenarios based on what has happened in the past, which is simply amazing.
In all cases, good analytics demands clear and good instructions from humans. It demands some grunt work from the data science teams, needing them to prepare the data, and point it to the analytics to perform its job, which encompasses building an appropriate set of rules for the analysis. This, however, poses our first problem – the human bias that gets embedded into the analytics. One wrong rule and your results might be far from the truth.
Yet again, relying purely on human assumptions and intuition doesn’t make your business grow either. I wrote in another article that you must let your data speak, and data speak usually through a data scientist or non-insightful BI systems. In the US alone – according to LinkedIn – there is a shortage of 150,000 data scientists. So, if you plan to rely on hiring people to bring your company to the AI world, it may not be easy. The CIO magazine calls it ‘the great data science shortage’. However, with Augmented Analytics the situation changes in many ways.
A simple solution? Adding artificial intelligence (AI) to the analytics.
AI can learn with the context – or data - and propose a possible set of actions or simply implement it automatically. It is also not static and can improve its own decisions with time. AI can learn from what the data covers, not necessarily from what humans programmed them to analyse.
Some executives still have some concerns about the use of AI, especially related to cyber threats, or giving space to AI to take wrong decisions. It potentially might force companies and individuals into compliance. These concerns prevent executives from relying more on these technologies, not allowing to serve their clients and businesses more efficiently.
However, data shows that not using AI is not helpful for businesses' performance.
Forrester has found that data-driven companies can grow eight times faster as compared to those that work from intuition. Meanwhile, Deloitte’s research shows that companies having high returns with AI are due to their high investments. It means that today, to achieve high results from AI you have to invest in it heavily. And to bring it to the banks' context, according to Deloitte, financial institutions are still in the quadrant of ‘low-investment/low-return’, which shows a big potential for improvement. Especially in operations’ optimization procedures, where data shows that current AI investments are proving profitable.
Currently, the big question for the executives is: how to access the high gains of AI without the huge investment it demands in teams and technologies?
The answer comes with Augmented Analytics (AA) technologies.
Data selection and preparation is one of the most significant processes in data science services. It takes time and a significant part of budgets, for instance, how to choose the right database? How to normalize it? How to aggregate and understand the consistency of the data? Often it takes a long time for good teams to do it – also often it comes with biases from the teams’ experience. It plays a special role when the data is unstructured or non-inspected, like in the process of natural language recognition (written or spoken). AA makes life extremely easy by eliminating all this grunt work and automatically selecting and preparing the data for you.
After the initial phase, AA automates the generation of insights. It is important to remark that, as the data reflects new customer behavior or changes in operations, insights tend to change as well, taking into account the new context.
It combines the best of the traditional Analytics and AI technologies – where Machine Learning (ML) is one of the most used technologies, not only to understand what's happening (context-sensitive) and what should be done (decision-making) but also to automate the process of doing it (self-learning).
Augmented analytics is not about the most updated technologies of AI. It is about using AI techniques to apply the analytics in companies - in scale - with less friction, less effort, fewer costs, and yet, achieving even higher results.
Some benefits of AA are:
It allows executives to have insights about their business without data scientists and IT teams;
It reduces manual labour and increases the effectiveness and efficiency of the IT and data science teams;
It increases the accuracy of the insights, adapting to new patterns and eliminating the biases of the previous analysis;
It substantially increases the speed and accuracy of the analysis, turning it into a continuous and automated process;
It uncovers hidden insights (costs and opportunities), especially the ones overlooked because of human bias;
It allows for improving the quality of the decision-making process.
Obviously, the move to a full AA environment is not easy. Traditionally IT departments are the owners of the databases and systems. It is still hard for executives of different levels – especially for CIOs – to accept that anyone could make queries into the databases about the business and have insights about it.
In the near future, not only companies and industries but also every department will be able to have insights based on real-time data, driven by Augmented Analytics embedded into their systems, for instance, CRM. However, it requires time – and governance. While this time hasn’t come yet, the transition approach today is to use AA to facilitate the effort of data scientists and IT teams to dramatically reduce costs, efforts and increase the accuracy of big-data projects. With a much more precise and efficient system, the AA tools are becoming game-changers for the companies’ financial performance and data consultancies of all kinds.
Pablo Morales | CEO of London Analytics Ltd.