Analytics is today providing a lot of decision-making power. There are five main types of analytics, each aimed at answering a different question:
- Descriptive – what is happening?
- Diagnostic – why did it happen?
- Discovery – what don’t I know that I should know?
- Prescriptive – what should I do about it?
- Predictive – what is likely to happen?
Big Data Analytics
Raw data, especially big data, offers very little value in its unprocessed state. Applying the right set of analytical tools, we can extract insights that can transform businesses.
The major focus since Big Data became the strongest buzzword in IT, around 2011, has been in capturing multiple streams of digital information. Corporate generated data has been supplemented with various sources of third party data, used to enrich the core corporate data.
To explain the differences between descriptive, predictive, discovery and prescriptive analytics we can look at it from the perspective of the value they each provide to the organisation.
With enough data, patterns emerge, upon which models to support the various types of analytics.
Descriptive analytics is the simplest class of analytics, reducing large volumes of data into contextually linked summaries of what has happened in the past. Roughly 80% of business analytics today is descriptive, largely due to the emergence of social analytics. For example, the number of fans, followers, page views, pins, tweets, posts, comments etc
Predictive analytics is the next step up in data processing. It uses a range of data mining, statistical modelling, and machine learning techniques to interrogate historical data, to make predictions about the future. It doesn’t aim to tell one what will happen in the future, rather it merely forecasts what might happen, with a defined degree of probability.
Sentiment analysis is a common type of predictive analytics: taking plain text and applying scoring algorithms to provide an output of either positive, negative, or something between “+1 and -1”.
Prescriptive analytics is a type of predictive analytics. However, it goes beyond descriptive and predictive models by recommending one or more courses of action – and showing the likely outcome of each decision. It is a powerful decision making tool – providing multiple versions of a future, based on various decision scenarios. It can also recommend the best course of action for any pre-specified outcome
Prescriptive analytics requires a predictive model with two additional components: actionable data and a feedback system that tracks the outcome produced by the action taken.