The importance and necessity of predictive analytics

With Big Data, comes Big Responsibility and even Greater Expectations.

Predictive analytics, a tranche of data analytics, uses historical and current data to make predictions about the conditions of a future occurrence. In doing so, organizations are able to predict the behaviour of their customers, asset managers can forecast asset values [1] and health professionals can predict the likelihood of a patient returning to hospital [2] – all more accurately and more efficiently than ever before. Corporations accordingly hold greater responsibility, towards themselves and consumers, to correctly map future behaviour, whilst consumers and clients in turn expect a higher level of predictive competence. In other words, “customers want every interaction to show you understand them” [3], their current desires and their future needs – predictive analytics makes this a very real possibility.

What are predictive analytics and how do they work?

The key tools involved in predictive analytics are machine learning, statistics and data mining [4]. Essentially, these technologies act to analyse datasets in order to detect patterns and create models that map out predictions. Popular approaches and methodologies are decision trees, neural networks and regression models [5]. When it comes to neural networks, the approach that makes use of machine learning, it is important that the training data is representative of the sort of data the model will make predictions on in the future [6]. In doing so, the model will learn from a viable sample and will therefore be able to make accurate predictions that are of use to the company in question. Regression models and decision trees more simply aim to map out patterns and decision journeys.

These predictive models, however, are only one of the forms that predictive analytics can take. Tools can also include the likes of anomaly detection and propensity scoring [7]. The latter involves ‘scoring’ customers on the probability of them engaging in a particular behaviour. In doing so, companies are able to identify which customers and which behaviours yield the greatest chance of conversion [8], a term used to refer to a customer engaging in a desired behaviour, such as making a purchase.

Whilst it’s true that we have long been able to draw conclusions on historical data through descriptive analytics, the importance of predictive analytics surrounds heightened capabilities brought about by machine learning that allow us to predict future outcomes. Machine learning tools are also able to spot trends in data that have previously gone undetected [9]. The potential held by predictive analytics is, therefore, hard to overstate.

How are predictive analytics being used?

Predictive analytics are being used across industries, from trading to retail and beyond. Ultimately, these tools enable companies to forecast movements, actions and behaviours and therefore provide insights that are of use to many.

McKinsey has highlighted the use of predictive analytics in areas as specific as product-development projects [10]. Given that the “underlying complexity drivers across projects are similar and can be quantified” [11], predictive analytics tools can forecast the timelines and resources needed for a given project – be it staffing, overall costs or physical supplies [12]. Accordingly, companies “can put the right number of the right people on their projects at the right time” [13], and therefore reduce the risk of overruns and ultimately lower associated costs [14]. As such, predictive analytics can greatly improve the overall efficiency of product-development projects.

Likewise, given that predictive analytics tools have the ability to map out consumer behaviour, Adobe has highlighted their usefulness in retargeting those tha