HomeData AnalyticsOperational intelligence, three ways – descriptive, predictive and prescriptive

Operational intelligence, three ways – descriptive, predictive and prescriptive

Data analytics, and the decision-making algorithms that run besides, is the single discipline that will make industry ‘smart’. Everything else – and arguably everything to this point – is just the squally preamble to the giant industrial storm these tools will bring.

The most transformative digital change will come with full-force data analytics, in the shape of advanced data processing and artificial intelligence (AI). In the meantime, the first gusts can be perceived already, as analytics is being applied to machine and business data.

In general, there are three ways to apply data analytics in enterprise operations: as descriptive analytics, predictive analytics, and prescriptive analytics, with sophistication and insight rising by turn. Here, we consider all three.


Descriptive analytics explains an outcome; it helps an enterprise understand why something has already happened, based on historical data. It is the simplest variety of analytics, reducing big data into small data, which can be translated and understood.

Most business analytics is descriptive; it underpins the dashboards available with most business insights tools, and web and social media programmes. The digital revolution just makes the capacity for descriptive analytics massive – crossing inputs from myriad data-sets into business insights.


Predictive analytics is the rage in enterprises, and notably industrial set-ups seeking to increase efficiencies by automating processes. Insights can be gleaned between the lines, to light up future scenarios.

In essence, data is carried forward to make estimates about the likelihood of future outcomes. The idea is to ‘predict’ what might happen in the future, as opposed to what will happen. No statistical algorithm can be certain, after all.

The calculation is about weighting probabilities – about using the available data to guess at the unavailable data.

It uses a variety of statistical, modelling, data mining, and machine learning techniques to study recent and historical data. Data is taken from any relevant system – from CRM, ERP, HR, and POS set-ups to more sector-specific functions. Relationships are captured between data sets, which enable calculated judgements about future scenarios.


Prescriptive analytics goes further than a straight prediction, by recommending a decision, or decisions, based on the future-gazing. It allows users to ‘prescribe’ actions to a probable outcome.

Prescriptive analytics requires actionable data and a feedback loop, to track the outcome produced by the action taken. The process seeks to give advice, and make a decision about a future path(s) based on the data journey so far.

It applies advanced algorithms, computational modelling, and machine learning to historical and transactional data, as well as real-time feeds. Prescriptive analytics is complex to administer; most companies are not yet using them in their daily course of business.

Previous post
Semtech spurs LoRa community with next-gen silicon, stares down cellular IoT carriers
Next post
NNNCo deploys IoT network for Australian cotton farmers